♻ ☆ FlowTracer: A Tool for Uncovering Network Path Usage Imbalance in AI Training Clusters

Hasibul Jamil, Abdul Alim, Laurent Schares, Pavlos Maniotis, Liran Schour, Ali Sydney, Abdullah Kayi, Tevfik Kosar, Bengi Karacali

The increasing complexity of AI workloads, especially distributed Large Language Model (LLM) training, places significant strain on the networking infrastructure of parallel data centers and supercomputing systems. While Equal-Cost Multi- Path (ECMP) routing distributes traffic over parallel paths, hash collisions often lead to imbalanced network resource utilization and performance bottlenecks. This paper presents FlowTracer, a tool designed to analyze network path utilization and evaluate different routing strategies. FlowTracer aids in debugging network inefficiencies by providing detailed visibility into traffic distribution and helping to identify the root causes of performance degradation, such as issues caused by hash collisions. By offering flow-level insights, FlowTracer enables system operators to optimize routing, reduce congestion, and improve the performance of distributed AI workloads. We use a RoCEv2-enabled cluster with a leaf-spine network and 16 400-Gbps nodes to demonstrate how FlowTracer can be used to compare the flow imbalances of ECMP routing against a statically configured network. The example showcases a 30% reduction in imbalance, as measured by a new metric we introduce.

comment: Submitted for peer reviewing in IEEE ICC 2025

♻ ☆ Efficient Scheduling of Vehicular Tasks on Edge Systems with Green Energy and Battery Storage

Suvarthi Sarkar, Abinash Kumar Ray, Aryabartta Sahu

The autonomous vehicle industry is rapidly expanding, requiring significant computational resources for tasks like perception and decision-making. Vehicular edge computing has emerged to meet this need, utilizing roadside computational units (roadside edge servers) to support autonomous vehicles. Aligning with the trend of green cloud computing, these roadside edge servers often get energy from solar power. Additionally, each roadside computational unit is equipped with a battery for storing solar power, ensuring continuous computational operation during periods of low solar energy availability. In our research, we address the scheduling of computational tasks generated by autonomous vehicles to roadside units with power consumption proportional to the cube of the computational load of the server. Each computational task is associated with a revenue, dependent on its computational needs and deadline. Our objective is to maximize the total revenue of the system of roadside computational units. We propose an offline heuristics approach based on predicted solar energy and incoming task patterns for different time slots. Additionally, we present heuristics for real-time adaptation to varying solar energy and task patterns from predicted values for different time slots. Our comparative analysis shows that our methods outperform state-of-the-art approaches upto 40\% for real-life datasets.

♻ ☆ PyGim: An Efficient Graph Neural Network Library for Real Processing-In-Memory Architectures

Christina Giannoula, Peiming Yang, Ivan Fernandez Vega, Jiacheng Yang, Sankeerth Durvasula, Yu Xin Li, Mohammad Sadrosadati, Juan Gomez Luna, Onur Mutlu, Gennady Pekhimenko

Graph Neural Networks (GNNs) are emerging ML models to analyze graph-structure data. Graph Neural Network (GNN) execution involves both compute-intensive and memory-intensive kernels, the latter dominates the total time, being significantly bottlenecked by data movement between memory and processors. Processing-In-Memory (PIM) systems can alleviate this data movement bottleneck by placing simple processors near or inside to memory arrays. In this work, we introduce PyGim, an efficient ML library that accelerates GNNs on real PIM systems. We propose intelligent parallelization techniques for memory-intensive kernels of GNNs tailored for real PIM systems, and develop handy Python API for them. We provide hybrid GNN execution, in which the compute-intensive and memory-intensive kernels are executed in processor-centric and memory-centric computing systems, respectively. We extensively evaluate PyGim on a real-world PIM system with 1992 PIM cores using emerging GNN models, and demonstrate that it outperforms its state-of-the-art CPU counterpart on Intel Xeon by on average 3.04x, and achieves higher resource utilization than CPU and GPU systems. Our work provides useful recommendations for software, system and hardware designers. PyGim is publicly available at https://github.com/CMU-SAFARI/PyGim.

♻ ☆ FlowTracer: A Tool for Uncovering Network Path Usage Imbalance in AI Training Clusters

Hasibul Jamil, Abdul Alim, Laurent Schares, Pavlos Maniotis, Liran Schour, Ali Sydney, Abdullah Kayi, Tevfik Kosar, Bengi Karacali

The increasing complexity of AI workloads, especially distributed Large Language Model (LLM) training, places significant strain on the networking infrastructure of parallel data centers and supercomputing systems. While Equal-Cost Multi- Path (ECMP) routing distributes traffic over parallel paths, hash collisions often lead to imbalanced network resource utilization and performance bottlenecks. This paper presents FlowTracer, a tool designed to analyze network path utilization and evaluate different routing strategies. FlowTracer aids in debugging network inefficiencies by providing detailed visibility into traffic distribution and helping to identify the root causes of performance degradation, such as issues caused by hash collisions. By offering flow-level insights, FlowTracer enables system operators to optimize routing, reduce congestion, and improve the performance of distributed AI workloads. We use a RoCEv2-enabled cluster with a leaf-spine network and 16 400-Gbps nodes to demonstrate how FlowTracer can be used to compare the flow imbalances of ECMP routing against a statically configured network. The example showcases a 30% reduction in imbalance, as measured by a new metric we introduce.

comment: Submitted for peer reviewing in IEEE ICC 2025

♻ ☆ Efficient Scheduling of Vehicular Tasks on Edge Systems with Green Energy and Battery Storage

Suvarthi Sarkar, Abinash Kumar Ray, Aryabartta Sahu

The autonomous vehicle industry is rapidly expanding, requiring significant computational resources for tasks like perception and decision-making. Vehicular edge computing has emerged to meet this need, utilizing roadside computational units (roadside edge servers) to support autonomous vehicles. Aligning with the trend of green cloud computing, these roadside edge servers often get energy from solar power. Additionally, each roadside computational unit is equipped with a battery for storing solar power, ensuring continuous computational operation during periods of low solar energy availability. In our research, we address the scheduling of computational tasks generated by autonomous vehicles to roadside units with power consumption proportional to the cube of the computational load of the server. Each computational task is associated with a revenue, dependent on its computational needs and deadline. Our objective is to maximize the total revenue of the system of roadside computational units. We propose an offline heuristics approach based on predicted solar energy and incoming task patterns for different time slots. Additionally, we present heuristics for real-time adaptation to varying solar energy and task patterns from predicted values for different time slots. Our comparative analysis shows that our methods outperform state-of-the-art approaches upto 40\% for real-life datasets.

♻ ☆ PyGim: An Efficient Graph Neural Network Library for Real Processing-In-Memory Architectures

Christina Giannoula, Peiming Yang, Ivan Fernandez Vega, Jiacheng Yang, Sankeerth Durvasula, Yu Xin Li, Mohammad Sadrosadati, Juan Gomez Luna, Onur Mutlu, Gennady Pekhimenko

Graph Neural Networks (GNNs) are emerging ML models to analyze graph-structure data. Graph Neural Network (GNN) execution involves both compute-intensive and memory-intensive kernels, the latter dominates the total time, being significantly bottlenecked by data movement between memory and processors. Processing-In-Memory (PIM) systems can alleviate this data movement bottleneck by placing simple processors near or inside to memory arrays. In this work, we introduce PyGim, an efficient ML library that accelerates GNNs on real PIM systems. We propose intelligent parallelization techniques for memory-intensive kernels of GNNs tailored for real PIM systems, and develop handy Python API for them. We provide hybrid GNN execution, in which the compute-intensive and memory-intensive kernels are executed in processor-centric and memory-centric computing systems, respectively. We extensively evaluate PyGim on a real-world PIM system with 1992 PIM cores using emerging GNN models, and demonstrate that it outperforms its state-of-the-art CPU counterpart on Intel Xeon by on average 3.04x, and achieves higher resource utilization than CPU and GPU systems. Our work provides useful recommendations for software, system and hardware designers. PyGim is publicly available at https://github.com/CMU-SAFARI/PyGim.

♻ ☆ Lightweight Correlation-Aware Table Compression NeurIPS 2024

Mihail Stoian, Alexander van Renen, Jan Kobiolka, Ping-Lin Kuo, Josif Grabocka, Andreas Kipf

The growing adoption of data lakes for managing relational data necessitates efficient, open storage formats that provide high scan performance and competitive compression ratios. While existing formats achieve fast scans through lightweight encoding techniques, they have reached a plateau in terms of minimizing storage footprint. Recently, correlation-aware compression schemes have been shown to reduce file sizes further. Yet, current approaches either incur significant scan overheads or require manual specification of correlations, limiting their practicability. We present $\texttt{Virtual}$, a framework that integrates seamlessly with existing open formats to automatically leverage data correlations, achieving substantial compression gains while having minimal scan performance overhead. Experiments on data-gov datasets show that $\texttt{Virtual}$ reduces file sizes by up to 40% compared to Apache Parquet.

comment: Third Table Representation Learning Workshop (TRL @ NeurIPS 2024)

♻ ☆ PyGim: An Efficient Graph Neural Network Library for Real Processing-In-Memory Architectures

Christina Giannoula, Peiming Yang, Ivan Fernandez Vega, Jiacheng Yang, Sankeerth Durvasula, Yu Xin Li, Mohammad Sadrosadati, Juan Gomez Luna, Onur Mutlu, Gennady Pekhimenko

Graph Neural Networks (GNNs) are emerging ML models to analyze graph-structure data. Graph Neural Network (GNN) execution involves both compute-intensive and memory-intensive kernels, the latter dominates the total time, being significantly bottlenecked by data movement between memory and processors. Processing-In-Memory (PIM) systems can alleviate this data movement bottleneck by placing simple processors near or inside to memory arrays. In this work, we introduce PyGim, an efficient ML library that accelerates GNNs on real PIM systems. We propose intelligent parallelization techniques for memory-intensive kernels of GNNs tailored for real PIM systems, and develop handy Python API for them. We provide hybrid GNN execution, in which the compute-intensive and memory-intensive kernels are executed in processor-centric and memory-centric computing systems, respectively. We extensively evaluate PyGim on a real-world PIM system with 1992 PIM cores using emerging GNN models, and demonstrate that it outperforms its state-of-the-art CPU counterpart on Intel Xeon by on average 3.04x, and achieves higher resource utilization than CPU and GPU systems. Our work provides useful recommendations for software, system and hardware designers. PyGim is publicly available at https://github.com/CMU-SAFARI/PyGim.

♻ ☆ PnLCalib: Sports Field Registration via Points and Lines Optimization

Marc Gutiérrez-Pérez, Antonio Agudo

Camera calibration in broadcast sports videos presents numerous challenges for accurate sports field registration due to multiple camera angles, varying camera parameters, and frequent occlusions of the field. Traditional search-based methods depend on initial camera pose estimates, which can struggle in non-standard positions and dynamic environments. In response, we propose an optimization-based calibration pipeline that leverages a 3D soccer field model and a predefined set of keypoints to overcome these limitations. Our method also introduces a novel refinement module that improves initial calibration by using detected field lines in a non-linear optimization process. This approach outperforms existing techniques in both multi-view and single-view 3D camera calibration tasks, while maintaining competitive performance in homography estimation. Extensive experimentation on real-world soccer datasets, including SoccerNet-Calibration, WorldCup 2014, and TS-WorldCup, highlights the robustness and accuracy of our method across diverse broadcast scenarios. Our approach offers significant improvements in camera calibration precision and reliability.

comment: Extended version of "No Bells, Just Whistles: Sports Field Registration Leveraging Geometric Properties"

♻ ☆ A Review of Prominent Paradigms for LLM-Based Agents: Tool Use (Including RAG), Planning, and Feedback Learning

Xinzhe Li

Tool use, planning, and feedback learning are currently three prominent paradigms for developing Large Language Model (LLM)-based agents across various tasks. Although numerous frameworks have been devised for each paradigm, their intricate workflows and inconsistent taxonomy create challenges in understanding and reviewing the frameworks across different paradigms. This survey introduces a unified taxonomy to systematically review and discuss these frameworks. Specifically, 1) the taxonomy defines environments/tasks, common LLM-profiled roles or LMPRs (policy models, evaluators, and dynamic models), and universally applicable workflows found in prior work, and 2) it enables a comparison of key perspectives on the implementations of LMPRs and workflow designs across different agent paradigms and frameworks. 3) Finally, we identify three limitations in existing workflow designs and systematically discuss the future work. Resources have been made publicly available at in our GitHub repository https://github.com/xinzhel/LLM-Agent-Survey.

comment: Under Review

♻ ☆ Bonsai: Gradient-free Graph Distillation for Node Classification

Mridul Gupta, Samyak Jain, Vansh Ramani, Hariprasad Kodamana, Sayan Ranu

Graph distillation has emerged as a promising avenue to enable scalable training of GNNs by compressing the training dataset while preserving essential graph characteristics. Our study uncovers significant shortcomings in current graph distillation techniques. First, the majority of the algorithms paradoxically require training on the full dataset to perform distillation. Second, due to their gradient-emulating approach, these methods require fresh distillation for any change in hyperparameters or GNN architecture, limiting their flexibility and reusability. Finally, they fail to achieve substantial size reduction due to synthesizing fully-connected, edge-weighted graphs. To address these challenges, we present Bonsai, a novel graph distillation method empowered by the observation that \textit{computation trees} form the fundamental processing units of message-passing GNNs. Bonsai distills datasets by encoding a careful selection of \textit{exemplar} trees that maximize the representation of all computation trees in the training set. This unique approach imparts Bonsai as the first linear-time, model-agnostic graph distillation algorithm for node classification that outperforms existing baselines across $6$ real-world datasets on accuracy, while being $22$ times faster on average. Bonsai is grounded in rigorous mathematical guarantees on the adopted approximation strategies making it robust to GNN architectures, datasets, and parameters.

♻ ☆ RealignDiff: Boosting Text-to-Image Diffusion Model with Coarse-to-fine Semantic Re-alignment

Zutao Jiang, Guian Fang, Jianhua Han, Guansong Lu, Hang Xu, Shengcai Liao, Xiaojun Chang, Xiaodan Liang

Recent advances in text-to-image diffusion models have achieved remarkable success in generating high-quality, realistic images from textual descriptions. However, these approaches have faced challenges in precisely aligning the generated visual content with the textual concepts described in the prompts. In this paper, we propose a two-stage coarse-to-fine semantic re-alignment method, named RealignDiff, aimed at improving the alignment between text and images in text-to-image diffusion models. In the coarse semantic re-alignment phase, a novel caption reward, leveraging the BLIP-2 model, is proposed to evaluate the semantic discrepancy between the generated image caption and the given text prompt. Subsequently, the fine semantic re-alignment stage employs a local dense caption generation module and a re-weighting attention modulation module to refine the previously generated images from a local semantic view. Experimental results on the MS-COCO and ViLG-300 datasets demonstrate that the proposed two-stage coarse-to-fine semantic re-alignment method outperforms other baseline re-alignment techniques by a substantial margin in both visual quality and semantic similarity with the input prompt.

♻ ☆ Advancing Interpretability in Text Classification through Prototype Learning

Bowen Wei, Ziwei Zhu

Deep neural networks have achieved remarkable performance in various text-based tasks but often lack interpretability, making them less suitable for applications where transparency is critical. To address this, we propose ProtoLens, a novel prototype-based model that provides fine-grained, sub-sentence level interpretability for text classification. ProtoLens uses a Prototype-aware Span Extraction module to identify relevant text spans associated with learned prototypes and a Prototype Alignment mechanism to ensure prototypes are semantically meaningful throughout training. By aligning the prototype embeddings with human-understandable examples, ProtoLens provides interpretable predictions while maintaining competitive accuracy. Extensive experiments demonstrate that ProtoLens outperforms both prototype-based and non-interpretable baselines on multiple text classification benchmarks. Code and data are available at \url{https://anonymous.4open.science/r/ProtoLens-CE0B/}.

♻ ☆ RegExplainer: Generating Explanations for Graph Neural Networks in Regression Tasks NeurIPS 2024

Jiaxing Zhang, Zhuomin Chen, Hao Mei, Longchao Da, Dongsheng Luo, Hua Wei

Graph regression is a fundamental task that has gained significant attention in various graph learning tasks. However, the inference process is often not easily interpretable. Current explanation techniques are limited to understanding Graph Neural Network (GNN) behaviors in classification tasks, leaving an explanation gap for graph regression models. In this work, we propose a novel explanation method to interpret the graph regression models (XAIG-R). Our method addresses the distribution shifting problem and continuously ordered decision boundary issues that hinder existing methods away from being applied in regression tasks. We introduce a novel objective based on the graph information bottleneck theory (GIB) and a new mix-up framework, which can support various GNNs and explainers in a model-agnostic manner. Additionally, we present a self-supervised learning strategy to tackle the continuously ordered labels in regression tasks. We evaluate our proposed method on three benchmark datasets and a real-life dataset introduced by us, and extensive experiments demonstrate its effectiveness in interpreting GNN models in regression tasks.

comment: Accepted by NeurIPS 2024

♻ ☆ Evaluating AI-Generated Essays with GRE Analytical Writing Assessment

Yang Zhong, Jiangang Hao, Michael Fauss, Chen Li, Yuan Wang

The recent revolutionary advance in generative AI enables the generation of realistic and coherent texts by large language models (LLMs). Despite many existing evaluation metrics on the quality of the generated texts, there is still a lack of rigorous assessment of how well LLMs perform in complex and demanding writing assessments. This study examines essays generated by ten leading LLMs for the analytical writing assessment of the Graduate Record Exam (GRE). We assessed these essays using both human raters and the e-rater automated scoring engine as used in the GRE scoring pipeline. Notably, the top-performing Gemini and GPT-4o received an average score of 4.78 and 4.67, respectively, falling between "generally thoughtful, well-developed analysis of the issue and conveys meaning clearly" and "presents a competent analysis of the issue and conveys meaning with acceptable clarity" according to the GRE scoring guideline. We also evaluated the detection accuracy of these essays, with detectors trained on essays generated by the same and different LLMs.

comment: 20 pages, 6 figures

♻ ☆ HyperspectralViTs: General Hyperspectral Models for On-board Remote Sensing

Vít Růžička, Andrew Markham

On-board processing of hyperspectral data with machine learning models would enable unprecedented amount of autonomy for a wide range of tasks, for example methane detection or mineral identification. This can enable early warning system and could allow new capabilities such as automated scheduling across constellations of satellites. Classical methods suffer from high false positive rates and previous deep learning models exhibit prohibitive computational requirements. We propose fast and accurate machine learning architectures which support end-to-end training with data of high spectral dimension without relying on hand-crafted products or spectral band compression preprocessing. We evaluate our models on two tasks related to hyperspectral data processing. With our proposed general architectures, we improve the F1 score of the previous methane detection state-of-the-art models by 27% on a newly created synthetic dataset and by 13% on the previously released large benchmark dataset. We also demonstrate that training models on the synthetic dataset improves performance of models finetuned on the dataset of real events by 6.9% in F1 score in contrast with training from scratch. On a newly created dataset for mineral identification, our models provide 3.5% improvement in the F1 score in contrast to the default versions of the models. With our proposed models we improve the inference speed by 85% in contrast to previous classical and deep learning approaches by removing the dependency on classically computed features. With our architecture, one capture from the EMIT sensor can be processed within 30 seconds on realistic proxy of the ION-SCV 004 satellite.

comment: 13 pages, This work has been submitted to the IEEE for possible publication

♻ ☆ Estimating the Spectral Moments of the Kernel Integral Operator from Finite Sample Matrices

Chanwoo Chun, SueYeon Chung, Daniel D. Lee

Analyzing the structure of sampled features from an input data distribution is challenging when constrained by limited measurements in both the number of inputs and features. Traditional approaches often rely on the eigenvalue spectrum of the sample covariance matrix derived from finite measurement matrices; however, these spectra are sensitive to the size of the measurement matrix, leading to biased insights. In this paper, we introduce a novel algorithm that provides unbiased estimates of the spectral moments of the kernel integral operator in the limit of infinite inputs and features from finitely sampled measurement matrices. Our method, based on dynamic programming, is efficient and capable of estimating the moments of the operator spectrum. We demonstrate the accuracy of our estimator on radial basis function (RBF) kernels, highlighting its consistency with the theoretical spectra. Furthermore, we showcase the practical utility and robustness of our method in understanding the geometry of learned representations in neural networks.

♻ ☆ Bonsai: Gradient-free Graph Distillation for Node Classification

Mridul Gupta, Samyak Jain, Vansh Ramani, Hariprasad Kodamana, Sayan Ranu

Graph distillation has emerged as a promising avenue to enable scalable training of GNNs by compressing the training dataset while preserving essential graph characteristics. Our study uncovers significant shortcomings in current graph distillation techniques. First, the majority of the algorithms paradoxically require training on the full dataset to perform distillation. Second, due to their gradient-emulating approach, these methods require fresh distillation for any change in hyperparameters or GNN architecture, limiting their flexibility and reusability. Finally, they fail to achieve substantial size reduction due to synthesizing fully-connected, edge-weighted graphs. To address these challenges, we present Bonsai, a novel graph distillation method empowered by the observation that \textit{computation trees} form the fundamental processing units of message-passing GNNs. Bonsai distills datasets by encoding a careful selection of \textit{exemplar} trees that maximize the representation of all computation trees in the training set. This unique approach imparts Bonsai as the first linear-time, model-agnostic graph distillation algorithm for node classification that outperforms existing baselines across $6$ real-world datasets on accuracy, while being $22$ times faster on average. Bonsai is grounded in rigorous mathematical guarantees on the adopted approximation strategies making it robust to GNN architectures, datasets, and parameters.

♻ ☆ RegExplainer: Generating Explanations for Graph Neural Networks in Regression Tasks NeurIPS 2024

Jiaxing Zhang, Zhuomin Chen, Hao Mei, Longchao Da, Dongsheng Luo, Hua Wei

Graph regression is a fundamental task that has gained significant attention in various graph learning tasks. However, the inference process is often not easily interpretable. Current explanation techniques are limited to understanding Graph Neural Network (GNN) behaviors in classification tasks, leaving an explanation gap for graph regression models. In this work, we propose a novel explanation method to interpret the graph regression models (XAIG-R). Our method addresses the distribution shifting problem and continuously ordered decision boundary issues that hinder existing methods away from being applied in regression tasks. We introduce a novel objective based on the graph information bottleneck theory (GIB) and a new mix-up framework, which can support various GNNs and explainers in a model-agnostic manner. Additionally, we present a self-supervised learning strategy to tackle the continuously ordered labels in regression tasks. We evaluate our proposed method on three benchmark datasets and a real-life dataset introduced by us, and extensive experiments demonstrate its effectiveness in interpreting GNN models in regression tasks.

comment: Accepted by NeurIPS 2024

♻ ☆ Understanding When Tree of Thoughts Succeeds: Larger Models Excel in Generation, Not Discrimination

Qiqi Chen, Xinpeng Wang, Philipp Mondorf, Michael A. Hedderich, Barbara Plank

Tree of Thoughts (ToT) is a reasoning strategy for Large Language Models (LLMs) that employs a generator to suggest reasoning steps and a discriminator to decide which steps to implement. ToT demonstrates strong performance on reasoning tasks, often surpassing simple methods such as Input-Output (IO) prompting and Chain-of-Thought (CoT) reasoning. However, ToT does not consistently outperform such simpler methods across all models, leaving large knowledge gaps on the conditions under which ToT is most beneficial. In this paper, we analyze the roles of the generator and discriminator separately to better understand the conditions when ToT is beneficial. We find that the generator plays a more critical role than the discriminator in driving the success of ToT. Scaling the generator leads to notable improvements in ToT performance, even when using a smaller model as the discriminator, whereas scaling the discriminator with a fixed generator yields only marginal gains. Our results show that models across different scales exhibit comparable discrimination capabilities, yet differ significantly in their generative performance for ToT.

comment: Code: github.com/mainlp/tot-eval

♻ ☆ A Review of Prominent Paradigms for LLM-Based Agents: Tool Use (Including RAG), Planning, and Feedback Learning

Xinzhe Li

Tool use, planning, and feedback learning are currently three prominent paradigms for developing Large Language Model (LLM)-based agents across various tasks. Although numerous frameworks have been devised for each paradigm, their intricate workflows and inconsistent taxonomy create challenges in understanding and reviewing the frameworks across different paradigms. This survey introduces a unified taxonomy to systematically review and discuss these frameworks. Specifically, 1) the taxonomy defines environments/tasks, common LLM-profiled roles or LMPRs (policy models, evaluators, and dynamic models), and universally applicable workflows found in prior work, and 2) it enables a comparison of key perspectives on the implementations of LMPRs and workflow designs across different agent paradigms and frameworks. 3) Finally, we identify three limitations in existing workflow designs and systematically discuss the future work. Resources have been made publicly available at in our GitHub repository https://github.com/xinzhel/LLM-Agent-Survey.

comment: Under Review

♻ ☆ ESpeW: Robust Copyright Protection for LLM-based EaaS via Embedding-Specific Watermark

Zongqi Wang, Baoyuan Wu, Jingyuan Deng, Yujiu Yang

Embeddings as a Service (EaaS) is emerging as a crucial role in AI applications. Unfortunately, EaaS is vulnerable to model extraction attacks, highlighting the urgent need for copyright protection. Although some preliminary works propose applying embedding watermarks to protect EaaS, recent research reveals that these watermarks can be easily removed. Hence, it is crucial to inject robust watermarks resistant to watermark removal attacks. Existing watermarking methods typically inject a target embedding into embeddings through linear interpolation when the text contains triggers. However, this mechanism results in each watermarked embedding having the same component, which makes the watermark easy to identify and eliminate. Motivated by this, in this paper, we propose a novel embedding-specific watermarking (ESpeW) mechanism to offer robust copyright protection for EaaS. Our approach involves injecting unique, yet readily identifiable watermarks into each embedding. Watermarks inserted by ESpeW are designed to maintain a significant distance from one another and to avoid sharing common components, thus making it significantly more challenging to remove the watermarks. Extensive experiments on four popular datasets demonstrate that ESpeW can even watermark successfully against a highly aggressive removal strategy without sacrificing the quality of embeddings. Code is available at https://github.com/liudan193/ESpeW.

♻ ☆ Advancing Interpretability in Text Classification through Prototype Learning

Bowen Wei, Ziwei Zhu

Deep neural networks have achieved remarkable performance in various text-based tasks but often lack interpretability, making them less suitable for applications where transparency is critical. To address this, we propose ProtoLens, a novel prototype-based model that provides fine-grained, sub-sentence level interpretability for text classification. ProtoLens uses a Prototype-aware Span Extraction module to identify relevant text spans associated with learned prototypes and a Prototype Alignment mechanism to ensure prototypes are semantically meaningful throughout training. By aligning the prototype embeddings with human-understandable examples, ProtoLens provides interpretable predictions while maintaining competitive accuracy. Extensive experiments demonstrate that ProtoLens outperforms both prototype-based and non-interpretable baselines on multiple text classification benchmarks. Code and data are available at \url{https://anonymous.4open.science/r/ProtoLens-CE0B/}.

♻ ☆ Interação entre robôs humanoides: desenvolvendo a colaboração e comunicação autônoma

Moraes Pablo, Rodríguez Mónica, Peters Christopher, Sodre Hiago, Mazondo Ahilen, Sandin Vincent, Barcelona Sebastian, Moraes William, Fernández Santiago, Assunção Nathalie, de Vargas Bruna, Dörnbach Tobias, Kelbouscas André, Grando Ricardo

This study investigates the interaction between humanoid robots NAO and Pepper, emphasizing their potential applications in educational settings. NAO, widely used in education, and Pepper, designed for social interactions, of er new opportunities for autonomous communication and collaboration. Through a series of programmed interactions, the robots demonstrated their ability to communicate and coordinate actions autonomously, highlighting their potential as tools for enhancing learning environments. The research also explores the integration of emerging technologies, such as artificial intelligence, into these systems, allowing robots to learn from each other and adapt their behavior. The findings suggest that NAO and Pepper can significantly contribute to both technical learning and the development of social and emotional skills in students, of ering innovative pedagogical approaches through the use of humanoid robotics.

comment: in Portuguese language

♻ ☆ Evaluating AI-Generated Essays with GRE Analytical Writing Assessment

Yang Zhong, Jiangang Hao, Michael Fauss, Chen Li, Yuan Wang

The recent revolutionary advance in generative AI enables the generation of realistic and coherent texts by large language models (LLMs). Despite many existing evaluation metrics on the quality of the generated texts, there is still a lack of rigorous assessment of how well LLMs perform in complex and demanding writing assessments. This study examines essays generated by ten leading LLMs for the analytical writing assessment of the Graduate Record Exam (GRE). We assessed these essays using both human raters and the e-rater automated scoring engine as used in the GRE scoring pipeline. Notably, the top-performing Gemini and GPT-4o received an average score of 4.78 and 4.67, respectively, falling between "generally thoughtful, well-developed analysis of the issue and conveys meaning clearly" and "presents a competent analysis of the issue and conveys meaning with acceptable clarity" according to the GRE scoring guideline. We also evaluated the detection accuracy of these essays, with detectors trained on essays generated by the same and different LLMs.

comment: 20 pages, 6 figures

♻ ☆ BRIGHT: A Realistic and Challenging Benchmark for Reasoning-Intensive Retrieval

Hongjin Su, Howard Yen, Mengzhou Xia, Weijia Shi, Niklas Muennighoff, Han-yu Wang, Haisu Liu, Quan Shi, Zachary S. Siegel, Michael Tang, Ruoxi Sun, Jinsung Yoon, Sercan O. Arik, Danqi Chen, Tao Yu

Existing retrieval benchmarks primarily consist of information-seeking queries (e.g., aggregated questions from search engines) where keyword or semantic-based retrieval is usually sufficient. However, many complex real-world queries require in-depth reasoning to identify relevant documents that go beyond surface form matching. For example, finding documentation for a coding question requires understanding the logic and syntax of the functions involved. To better benchmark retrieval on such challenging queries, we introduce BRIGHT, the first text retrieval benchmark that requires intensive reasoning to retrieve relevant documents. Our dataset consists of 1,384 real-world queries spanning diverse domains, such as economics, psychology, mathematics, and coding. These queries are drawn from naturally occurring and carefully curated human data. Extensive evaluation reveals that even state-of-the-art retrieval models perform poorly on BRIGHT. The leading model on the MTEB leaderboard (Muennighoff et al., 2023), which achieves a score of 59.0 nDCG@10, produces a score of nDCG@10 of 18.3 on BRIGHT. We show that incorporating explicit reasoning about the query improves retrieval performance by up to 12.2 points. Moreover, incorporating retrieved documents from the top-performing retriever boosts question-answering performance by over 6.6 points. We believe that BRIGHT paves the way for future research on retrieval systems in more realistic and challenging settings.

comment: 48 pages

♻ ☆ Learning Mathematical Rules with Large Language Models NeurIPS'24

Nelson Vadori, Antoine Gorceix, Bastien Le Chenadec, Ahmad Rammal, Manuela Veloso

In this paper, we study the ability of large language models to learn specific mathematical rules such as distributivity or simplifying equations. We present an empirical analysis of their ability to generalize these rules, as well as to reuse them in the context of word problems. For this purpose, we provide a rigorous methodology to build synthetic data incorporating such rules, and perform fine-tuning of large language models on such data. Our experiments show that our model can learn and generalize these rules to some extent, as well as suitably reuse them in the context of word problems.

comment: NeurIPS'24 MATH-AI, the 4th Workshop on Mathematical Reasoning and AI

♻ ☆ System 2 thinking in OpenAI's o1-preview model: Near-perfect performance on a mathematics exam

Joost de Winter, Dimitra Dodou, Yke Bauke Eisma

The processes underlying human cognition are often divided into System 1, which involves fast, intuitive thinking, and System 2, which involves slow, deliberate reasoning. Previously, large language models were criticized for lacking the deeper, more analytical capabilities of System 2. In September 2024, OpenAI introduced the o1 model series, designed to handle System 2-like reasoning. While OpenAI's benchmarks are promising, independent validation is still needed. In this study, we tested the o1-preview model twice on the Dutch 'Mathematics B' final exam. It scored a near-perfect 76 and 74 out of 76 points. For context, only 24 out of 16,414 students in the Netherlands achieved a perfect score. By comparison, the GPT-4o model scored 66 and 62 out of 76, well above the Dutch students' average of 40.63 points. Neither model had access to the exam figures. Since there was a risk of model contami-nation (i.e., the knowledge cutoff for o1-preview and GPT-4o was after the exam was published online), we repeated the procedure with a new Mathematics B exam that was published after the cutoff date. The results again indicated that o1-preview performed strongly (97.8th percentile), which suggests that contamination was not a factor. We also show that there is some variability in the output of o1-preview, which means that sometimes there is 'luck' (the answer is correct) or 'bad luck' (the output has diverged into something that is incorrect). We demonstrate that the self-consistency approach, where repeated prompts are given and the most common answer is selected, is a useful strategy for identifying the correct answer. It is concluded that while OpenAI's new model series holds great potential, certain risks must be considered.

♻ ☆ PnLCalib: Sports Field Registration via Points and Lines Optimization

Marc Gutiérrez-Pérez, Antonio Agudo

Camera calibration in broadcast sports videos presents numerous challenges for accurate sports field registration due to multiple camera angles, varying camera parameters, and frequent occlusions of the field. Traditional search-based methods depend on initial camera pose estimates, which can struggle in non-standard positions and dynamic environments. In response, we propose an optimization-based calibration pipeline that leverages a 3D soccer field model and a predefined set of keypoints to overcome these limitations. Our method also introduces a novel refinement module that improves initial calibration by using detected field lines in a non-linear optimization process. This approach outperforms existing techniques in both multi-view and single-view 3D camera calibration tasks, while maintaining competitive performance in homography estimation. Extensive experimentation on real-world soccer datasets, including SoccerNet-Calibration, WorldCup 2014, and TS-WorldCup, highlights the robustness and accuracy of our method across diverse broadcast scenarios. Our approach offers significant improvements in camera calibration precision and reliability.

comment: Extended version of "No Bells, Just Whistles: Sports Field Registration Leveraging Geometric Properties"

♻ ☆ ERX: A Fast Real-Time Anomaly Detection Algorithm for Hyperspectral Line Scanning

Samuel Garske, Bradley Evans, Christopher Artlett, KC Wong

Detecting unexpected objects (anomalies) in real time has great potential for monitoring, managing, and protecting the environment. Hyperspectral line-scan cameras are a low-cost solution that enhance confidence in anomaly detection over RGB and multispectral imagery. However, existing line-scan algorithms are too slow when using small computers (e.g. those onboard a drone or small satellite), do not adapt to changing scenery, or lack robustness against geometric distortions. This paper introduces the Exponentially moving RX algorithm (ERX) to address these issues, and compares it with existing RX-based anomaly detection methods for hyperspectral line scanning. Three large and more complex datasets are also introduced to better assess the practical challenges when using line-scan cameras (two hyperspectral and one multispectral). ERX is evaluated using a Jetson Xavier NX compute module, achieving the best combination of speed and detection performance. This research paves the way for future studies in grouping and locating anomalous objects, adaptive and automatic threshold selection, and real-time field tests. The datasets and the Python code are available at: https://github.com/WiseGamgee/HyperAD.

comment: 17 pages, 13 figures, 4 tables, code and datasets accessible at https://github.com/WiseGamgee/HyperAD

♻ ☆ RealignDiff: Boosting Text-to-Image Diffusion Model with Coarse-to-fine Semantic Re-alignment

Zutao Jiang, Guian Fang, Jianhua Han, Guansong Lu, Hang Xu, Shengcai Liao, Xiaojun Chang, Xiaodan Liang

Recent advances in text-to-image diffusion models have achieved remarkable success in generating high-quality, realistic images from textual descriptions. However, these approaches have faced challenges in precisely aligning the generated visual content with the textual concepts described in the prompts. In this paper, we propose a two-stage coarse-to-fine semantic re-alignment method, named RealignDiff, aimed at improving the alignment between text and images in text-to-image diffusion models. In the coarse semantic re-alignment phase, a novel caption reward, leveraging the BLIP-2 model, is proposed to evaluate the semantic discrepancy between the generated image caption and the given text prompt. Subsequently, the fine semantic re-alignment stage employs a local dense caption generation module and a re-weighting attention modulation module to refine the previously generated images from a local semantic view. Experimental results on the MS-COCO and ViLG-300 datasets demonstrate that the proposed two-stage coarse-to-fine semantic re-alignment method outperforms other baseline re-alignment techniques by a substantial margin in both visual quality and semantic similarity with the input prompt.

♻ ☆ ARBEx: Attentive Feature Extraction with Reliability Balancing for Robust Facial Expression Learning ACCV 2024

Azmine Toushik Wasi, Karlo Šerbetar, Raima Islam, Taki Hasan Rafi, Dong-Kyu Chae

In this paper, we introduce a framework ARBEx, a novel attentive feature extraction framework driven by Vision Transformer with reliability balancing to cope against poor class distributions, bias, and uncertainty in the facial expression learning (FEL) task. We reinforce several data pre-processing and refinement methods along with a window-based cross-attention ViT to squeeze the best of the data. We also employ learnable anchor points in the embedding space with label distributions and multi-head self-attention mechanism to optimize performance against weak predictions with reliability balancing, which is a strategy that leverages anchor points, attention scores, and confidence values to enhance the resilience of label predictions. To ensure correct label classification and improve the models' discriminative power, we introduce anchor loss, which encourages large margins between anchor points. Additionally, the multi-head self-attention mechanism, which is also trainable, plays an integral role in identifying accurate labels. This approach provides critical elements for improving the reliability of predictions and has a substantial positive effect on final prediction capabilities. Our adaptive model can be integrated with any deep neural network to forestall challenges in various recognition tasks. Our strategy outperforms current state-of-the-art methodologies, according to extensive experiments conducted in a variety of contexts.

comment: Extended version is accepted in ACCV 2024 as GReFEL

♻ ☆ ReCAP: Recursive Cross Attention Network for Pseudo-Label Generation in Robotic Surgical Skill Assessment

Julien Quarez, Marc Modat, Sebastien Ourselin, Jonathan Shapey, Alejandro Granados

In surgical skill assessment, the Objective Structured Assessments of Technical Skills (OSATS) and Global Rating Scale (GRS) are well-established tools for evaluating surgeons during training. These metrics, along with performance feedback, help surgeons improve and reach practice standards. Recent research on the open-source JIGSAWS dataset, which includes both GRS and OSATS labels, has focused on regressing GRS scores from kinematic data, video, or their combination. However, we argue that regressing GRS alone is limiting, as it aggregates OSATS scores and overlooks clinically meaningful variations during a surgical trial. To address this, we developed a recurrent transformer model that tracks a surgeon's performance throughout a session by mapping hidden states to six OSATS, derived from kinematic data, using a clinically motivated objective function. These OSATS scores are averaged to predict GRS, allowing us to compare our model's performance against state-of-the-art (SOTA) methods. We report Spearman's Correlation Coefficients (SCC) demonstrating that our model outperforms SOTA using kinematic data (SCC 0.83-0.88), and matches performance with video-based models. Our model also surpasses SOTA in most tasks for average OSATS predictions (SCC 0.46-0.70) and specific OSATS (SCC 0.56-0.95). The generation of pseudo-labels at the segment level translates quantitative predictions into qualitative feedback, vital for automated surgical skill assessment pipelines. A senior surgeon validated our model's outputs, agreeing with 77% of the weakly-supervised predictions (p=0.006).

♻ ☆ Hierarchical Multi-agent Reinforcement Learning for Cyber Network Defense AAMAS

Aditya Vikram Singh, Ethan Rathbun, Emma Graham, Lisa Oakley, Simona Boboila, Alina Oprea, Peter Chin

Recent advances in multi-agent reinforcement learning (MARL) have created opportunities to solve complex real-world tasks. Cybersecurity is a notable application area, where defending networks against sophisticated adversaries remains a challenging task typically performed by teams of security operators. In this work, we explore novel MARL strategies for building autonomous cyber network defenses that address challenges such as large policy spaces, partial observability, and stealthy, deceptive adversarial strategies. To facilitate efficient and generalized learning, we propose a hierarchical Proximal Policy Optimization (PPO) architecture that decomposes the cyber defense task into specific sub-tasks like network investigation and host recovery. Our approach involves training sub-policies for each sub-task using PPO enhanced with domain expertise. These sub-policies are then leveraged by a master defense policy that coordinates their selection to solve complex network defense tasks. Furthermore, the sub-policies can be fine-tuned and transferred with minimal cost to defend against shifts in adversarial behavior or changes in network settings. We conduct extensive experiments using CybORG Cage 4, the state-of-the-art MARL environment for cyber defense. Comparisons with multiple baselines across different adversaries show that our hierarchical learning approach achieves top performance in terms of convergence speed, episodic return, and several interpretable metrics relevant to cybersecurity, including the fraction of clean machines on the network, precision, and false positives on recoveries.

comment: 9 pages, 7 figures, AAMAS preprint

♻ ☆ Enhancing MOTION2NX for Efficient, Scalable and Secure Image Inference using Convolutional Neural Networks

Haritha K, Ramya Burra, Srishti Mittal, Sarthak Sharma, Abhilash Venkatesh, Anshoo Tandon

This work contributes towards the development of an efficient and scalable open-source Secure Multi-Party Computation (SMPC) protocol on machines with moderate computational resources. We use the ABY2.0 SMPC protocol implemented on the C++ based MOTION2NX framework for secure convolutional neural network (CNN) inference application with semi-honest security. Our list of contributions are as follows. Firstly, we enhance MOTION2NX by providing a tensorized version of several primitive functions including the Hadamard product, indicator function and argmax function. Secondly, we adapt an existing Helper node algorithm, working in tandem with the ABY2.0 protocol, for efficient convolution computation to reduce execution time and RAM usage. Thirdly, we also present a novel splitting algorithm that divides the computations at each CNN layer into multiple configurable chunks. This novel splitting algorithm, providing significant reduction in RAM usage, is of independent interest and is applicable to general SMPC protocols.

comment: 20 pages, 1 figure. arXiv admin note: text overlap with arXiv:2310.10133

♻ ☆ CountCrypt: Quantum Cryptography between QCMA and PP

Eli Goldin, Tomoyuki Morimae, Saachi Mutreja, Takashi Yamakawa

We construct a quantum oracle relative to which BQP = QCMA but quantum-computation-classical-communication (QCCC) key exchange, QCCC commitments, and two-round quantum key distribution exist. We also construct an oracle relative to which BQP = QMA, but quantum lightning (a stronger variant of quantum money) exists. This extends previous work by Kretschmer [Kretschmer, TQC22], which showed that there is a quantum oracle relative to which BQP = QMA but pseudorandom state generators (a quantum variant of pseudorandom generators) exist. We also show that QCCC key exchange, QCCC commitments, and two-round quantum key distribution can all be used to build one-way puzzles. One-way puzzles are a version of "quantum samplable" one-wayness and are an intermediate primitive between pseudorandom state generators and EFI pairs, the minimal quantum primitive. In particular, one-way puzzles cannot exist if BQP = PP. Our results together imply that aside from pseudorandom state generators, there is a large class of quantum cryptographic primitives which can exist even if BQP = QCMA, but are broken if BQP = PP. Furthermore, one-way puzzles are a minimal primitive for this class. We denote this class "CountCrypt".

comment: 50 pages, 1 figure

♻ ☆ FlowTracer: A Tool for Uncovering Network Path Usage Imbalance in AI Training Clusters

Hasibul Jamil, Abdul Alim, Laurent Schares, Pavlos Maniotis, Liran Schour, Ali Sydney, Abdullah Kayi, Tevfik Kosar, Bengi Karacali

The increasing complexity of AI workloads, especially distributed Large Language Model (LLM) training, places significant strain on the networking infrastructure of parallel data centers and supercomputing systems. While Equal-Cost Multi- Path (ECMP) routing distributes traffic over parallel paths, hash collisions often lead to imbalanced network resource utilization and performance bottlenecks. This paper presents FlowTracer, a tool designed to analyze network path utilization and evaluate different routing strategies. FlowTracer aids in debugging network inefficiencies by providing detailed visibility into traffic distribution and helping to identify the root causes of performance degradation, such as issues caused by hash collisions. By offering flow-level insights, FlowTracer enables system operators to optimize routing, reduce congestion, and improve the performance of distributed AI workloads. We use a RoCEv2-enabled cluster with a leaf-spine network and 16 400-Gbps nodes to demonstrate how FlowTracer can be used to compare the flow imbalances of ECMP routing against a statically configured network. The example showcases a 30% reduction in imbalance, as measured by a new metric we introduce.

comment: Submitted for peer reviewing in IEEE ICC 2025

♻ ☆ R-ACP: Real-Time Adaptive Collaborative Perception Leveraging Robust Task-Oriented Communications

Zhengru Fang, Jingjing Wang, Yanan Ma, Yihang Tao, Yiqin Deng, Xianhao Chen, Yuguang Fang

Collaborative perception enhances sensing in multi-robot and vehicular networks by fusing information from multiple agents, improving perception accuracy and sensing range. However, mobility and non-rigid sensor mounts introduce extrinsic calibration errors, necessitating online calibration, further complicated by limited overlap in sensing regions. Moreover, maintaining fresh information is crucial for timely and accurate sensing. To address calibration errors and ensure timely and accurate perception, we propose a robust task-oriented communication strategy to optimize online self-calibration and efficient feature sharing for Real-time Adaptive Collaborative Perception (R-ACP). Specifically, we first formulate an Age of Perceived Targets (AoPT) minimization problem to capture data timeliness of multi-view streaming. Then, in the calibration phase, we introduce a channel-aware self-calibration technique based on re-identification (Re-ID), which adaptively compresses key features according to channel capacities, effectively addressing calibration issues via spatial and temporal cross-camera correlations. In the streaming phase, we tackle the trade-off between bandwidth and inference accuracy by leveraging an Information Bottleneck (IB) based encoding method to adjust video compression rates based on task relevance, thereby reducing communication overhead and latency. Finally, we design a priority-aware network to filter corrupted features to mitigate performance degradation from packet corruption. Extensive studies demonstrate that our framework outperforms five baselines, improving multiple object detection accuracy (MODA) by 25.49% and reducing communication costs by 51.36% under severely poor channel conditions.

♻ ☆ The Neuromorphic Analog Electronic Nose

Shavika Rastogi, Nik Dennler, Michael Schmuker, André van Schaik

Rapid detection of gas concentration is important in different domains like gas leakage monitoring, pollution control, and so on, for the prevention of health hazards. Out of different types of gas sensors, Metal oxide (MOx) sensors are extensively used in such applications because of their portability, low cost, and high sensitivity for specific gases. However, how to effectively sample the MOx data for the real-time detection of gas and its concentration level remains an open question. Here we introduce a simple analog front-end for one MOx sensor that encodes the gas concentration in the time difference between pulses of two separate pathways. This front-end design is inspired by the spiking output of a mammalian olfactory bulb. We show that for a gas pulse injected in a constant airflow, the time difference between pulses decreases with increasing gas concentration, similar to the spike time difference between the two principal output neurons in the olfactory bulb. The circuit design is further extended to a MOx sensor array and this sensor array front-end was tested in the same environment for gas identification and concentration estimation. Encoding of gas stimulus features in analog spikes at the sensor level itself may result in data and power-efficient real-time gas sensing systems in the future that can ultimately be used in uncontrolled and turbulent environments for longer periods without data explosion.

☆ POD-Attention: Unlocking Full Prefill-Decode Overlap for Faster LLM Inference

Aditya K Kamath, Ramya Prabhu, Jayashree Mohan, Simon Peter, Ramachandran Ramjee, Ashish Panwar

Each request in LLM inference goes through two phases: compute-bound prefill and memory-bandwidth-bound decode. To improve GPU utilization, recent systems use hybrid batching that combines the prefill and decode phases of different requests into the same batch. Hybrid batching works well for linear operations as it amortizes the cost of loading model weights from HBM. However, attention computation in hybrid batches remains inefficient because existing attention kernels are optimized for either prefill or decode. In this paper, we present POD-Attention -- the first GPU kernel that efficiently computes attention for hybrid batches. POD-Attention aims to maximize the utilization of both compute and memory bandwidth by carefully allocating the GPU's resources such that prefill and decode operations happen concurrently on the same multiprocessor. We integrate POD-Attention in a state-of-the-art LLM inference scheduler Sarathi-Serve. POD-Attention speeds up attention computation by up to 75% (mean 28%) and increases LLM serving throughput by up to 22% in offline inference. In online inference, POD-Attention enables lower time-to-first-token (TTFT), time-between-tokens (TBT), and request execution latency versus Sarathi-Serve.

☆ Optimal Fault-Tolerant Dispersion on Oriented Grids

Rik Banerjee, Manish Kumar, Anisur Rahaman Molla

Dispersion of mobile robots over the nodes of an anonymous graph is an important problem and turns out to be a crucial subroutine for designing efficient algorithms for many fundamental graph problems via mobile robots. In this problem, starting from an arbitrary initial distribution of $n$ robots across the $n$ nodes, the goal is to achieve a final configuration where each node holds at most one robot. This paper investigates the dispersion problem on an oriented grid, considering the possibility of robot failures (crashes) at any time during the algorithm's execution. We present a crash-tolerant dispersion algorithm that solves the dispersion problem on an anonymous oriented grid in $O(\sqrt{n})$ time and using $O(\log n)$ bits of memory per robot. The algorithm is optimal in terms of both time and memory per robot. We further extend this algorithm to deal with weak Byzantine robots. The weak Byzantine fault dispersion algorithm takes optimal $O(\sqrt{n})$ rounds but requires $O(n\log n)$ bits of memory per robot.

comment: arXiv admin note: substantial text overlap with arXiv:2405.02002

☆ Deoxys: A Causal Inference Engine for Unhealthy Node Mitigation in Large-scale Cloud Infrastructure

Chaoyun Zhang, Randolph Yao, Si Qin, Ze Li, Shekhar Agrawal, Binit R. Mishra, Tri Tran, Minghua Ma, Qingwei Lin, Murali Chintalapati, Dongmei Zhang

The presence of unhealthy nodes in cloud infrastructure signals the potential failure of machines, which can significantly impact the availability and reliability of cloud services, resulting in negative customer experiences. Effectively addressing unhealthy node mitigation is therefore vital for sustaining cloud system performance. This paper introduces Deoxys, a causal inference engine tailored to recommending mitigation actions for unhealthy node in cloud systems to minimize virtual machine downtime and interruptions during unhealthy events. It employs double machine learning combined with causal forest to produce precise and reliable mitigation recommendations based solely on limited observational data collected from the historical unhealthy events. To enhance the causal inference model, Deoxys further incorporates a policy fallback mechanism based on model uncertainty and action overriding mechanisms to (i) improve the reliability of the system, and (ii) strike a good tradeoff between downtime reduction and resource utilization, thereby enhancing the overall system performance. After deploying Deoxys in a large-scale cloud infrastructure at Microsoft, our observations demonstrate that Deoxys significantly reduces average VM downtime by 53% compared to a legacy policy, while leading to 49.5% lower VM interruption rate. This substantial improvement enhances the reliability and stability of cloud platforms, resulting in a seamless customer experience.

☆ Signature-based IaaS Performance Change Detection

Sheik Mohammad Mostakim Fattah, Athman Bouguettaya

We propose a novel change detection framework to identify changes in the long-term performance behavior of an IaaS service. An IaaS service's long-term performance behavior is represented by an IaaS performance signature. The proposed framework leverages time series similarity measures and a sliding window technique to detect changes in IaaS performance signatures. We introduce a new IaaS performance noise model that enables the proposed framework to distinguish between performance noise and actual changes in performance. The proposed framework utilizes a novel Signal-to-Noise Ratio (SNR) based approach to detect changes when prior knowledge about performance noise is available. A set of experiments is conducted using real-world datasets to demonstrate the effectiveness of the proposed change detection framework.

comment: accepted in ACM transaction on Internet Technology in October 2024. arXiv admin note: text overlap with arXiv:2007.11705

☆ Efficiently Scheduling Parallel DAG Tasks on Identical Multiprocessors

Shardul Lendve, Konstantinos Bletsas, Pedro F. Souto

Parallel real-time embedded applications can be modelled as directed acyclic graphs (DAGs) whose nodes model subtasks and whose edges model precedence constraints among subtasks. Efficiently scheduling such parallel tasks can be challenging in itself, particularly in hard real-time systems where it must be ensured offline that the deadlines of the parallel applications will be met at run time. In this paper, we tackle the problem of scheduling DAG tasks on identical multiprocessor systems efficiently, in terms of processor utilisation. We propose a new algorithm that attempts to use dedicated processor clusters for high-utilisation tasks, as in federated scheduling, but is also capable of reclaiming the processing capacity lost to fragmentation, by splitting the execution of parallel tasks over different existing clusters, in a manner inspired by semi-partitioned C=D scheduling (originally devised for non-parallel tasks). In the experiments with synthetic DAG task sets, our Segmented-Flattened-and-Split scheduling approach achieves a significantly higher scheduling success ratio than federated scheduling.

comment: Version submitted to RTNS 2024, on 16/08/2024 (with some typos fixed)

♻ ☆ TiMePReSt: Time and Memory Efficient Pipeline Parallel DNN Training with Removed Staleness

Ankita Dutta, Nabendu Chaki, Rajat K. De

DNN training is time-consuming and requires efficient multi-accelerator parallelization, where a single training iteration is split over available accelerators. Current approaches often parallelize training using intra-batch parallelization. Combining inter-batch and intra-batch pipeline parallelism is common to further improve training throughput. In this article, we develop a system, called TiMePReSt, that combines them in a novel way which helps to better overlap computation and communication, and limits the amount of communication. The traditional pipeline-parallel training of DNNs maintains similar working principle as sequential or conventional training of DNNs by maintaining consistent weight versions in forward and backward passes of a mini-batch. Thus, it suffers from high GPU memory footprint during training. In this paper, experimental study demonstrates that compromising weight consistency doesn't decrease prediction capability of a parallelly trained DNN. Moreover, TiMePReSt overcomes GPU memory overhead and achieves zero weight staleness. State-of-the-art techniques often become costly in terms of training time. In order to address this issue, TiMePReSt introduces a variant of intra-batch parallelism that parallelizes the forward pass of each mini-batch by decomposing it into smaller micro-batches. A novel synchronization method between forward and backward passes reduces training time in TiMePReSt. The occurrence of multiple sequence problem and its relation with version difference have been observed in TiMePReSt. This paper presents a mathematical relationship between the number of micro-batches and worker machines, highlighting the variation in version difference. A mathematical expression has been developed to calculate version differences for various combinations of these two without creating diagrams for all combinations.

♻ ☆ Harnessing Your DRAM and SSD for Sustainable and Accessible LLM Inference with Mixed-Precision and Multi-level Caching

Jie Peng, Zhang Cao, Huaizhi Qu, Zhengyu Zhang, Chang Guo, Yanyong Zhang, Zhichao Cao, Tianlong Chen

Although Large Language Models (LLMs) have demonstrated remarkable capabilities, their massive parameter counts and associated extensive computing make LLMs' deployment the main part of carbon emission from nowadays AI applications. Compared to modern GPUs like H$100$, it would be significantly carbon-sustainable if we could leverage old-fashioned GPUs such as M$40$ (as shown in Figure 1, M$40$ only has one third carbon emission of H$100$'s) for LLM servings. However, the limited High Bandwidth Memory (HBM) available on such GPU often cannot support the loading of LLMs due to the gigantic model size and intermediate activation data, making their serving challenging. For instance, a LLaMA2 model with $70$B parameters typically requires $128$GB for inference, which substantially surpasses $24$GB HBM in a $3090$ GPU and remains infeasible even considering the additional $64$GB DRAM. To address this challenge, this paper proposes a mixed-precision with a model modularization algorithm to enable LLM inference on outdated hardware with resource constraints. (The precision denotes the numerical precision like FP16, INT8, INT4) and multi-level caching (M2Cache).) Specifically, our M2Cache first modulizes neurons in LLM and creates their importance ranking. Then, it adopts a dynamic sparse mixed-precision quantization mechanism in weight space to reduce computational demands and communication overhead at each decoding step. It collectively lowers the operational carbon emissions associated with LLM inference. Moreover, M2Cache introduces a three-level cache management system with HBM, DRAM, and SSDs that complements the dynamic sparse mixed-precision inference. To enhance communication efficiency, M2Cache maintains a neuron-level mixed-precision LRU cache in HBM, a larger layer-aware cache in DRAM, and a full model in SSD.

comment: 24 pages, 13 figures

♻ ☆ Large Scale Multi-GPU Based Parallel Traffic Simulation for Accelerated Traffic Assignment and Propagation

Xuan Jiang, Raja Sengupta, James Demmel, Samuel Williams

Traffic propagation simulation is crucial for urban planning, enabling congestion analysis, travel time estimation, and route optimization. Traditional micro-simulation frameworks are limited to main roads due to the complexity of urban mobility and large-scale data. We introduce the Large Scale Multi-GPU Parallel Computing based Regional Scale Traffic Simulation Framework (LPSim), a scalable tool that leverages GPU parallel computing to simulate extensive traffic networks with high fidelity and reduced computation time. LPSim performs millions of vehicle dynamics simulations simultaneously, outperforming CPU-based methods. It can complete simulations of 2.82 million trips in 6.28 minutes using a single GPU, and 9.01 million trips in 21.16 minutes on dual GPUs. LPSim is also tested on dual NVIDIA A100 GPUs, achieving simulations about 113 times faster than traditional CPU methods. This demonstrates its scalability and efficiency for large-scale applications, making LPSim a valuable resource for researchers and planners. Code: https://github.com/Xuan-1998/LPSim

♻ ☆ Cold Start Latency in Serverless Computing: A Systematic Review, Taxonomy, and Future Directions

Muhammed Golec, Guneet Kaur Walia, Mohit Kumar, Felix Cuadrado, Sukhpal Singh Gill, Steve Uhlig

Recently, academics and the corporate sector have paid attention to serverless computing, which enables dynamic scalability and an economic model. In serverless computing, users only pay for the time they actually use resources, enabling zero scaling to optimise cost and resource utilisation. However, this approach also introduces the serverless cold start problem. Researchers have developed various solutions to address the cold start problem, yet it remains an unresolved research area. In this article, we propose a systematic literature review on clod start latency in serverless computing. Furthermore, we create a detailed taxonomy of approaches to cold start latency, which we use to investigate existing techniques for reducing the cold start time and frequency. We have classified the current studies on cold start latency into several categories such as caching and application-level optimisation-based solutions, as well as Artificial Intelligence (AI)/Machine Learning (ML)-based solutions. Moreover, we have analyzed the impact of cold start latency on quality of service, explored current cold start latency mitigation methods, datasets, and implementation platforms, and classified them into categories based on their common characteristics and features. Finally, we outline the open challenges and highlight the possible future directions.

comment: Preprint Version Accepted for Publication in ACM Computing Survey, 2024

♻ ☆ I've Got 99 Problems But FLOPS Ain't One

Alexandru M. Gherghescu, Vlad-Andrei Bădoiu, Alexandru Agache, Mihai-Valentin Dumitru, Iuliu Vasilescu, Radu Mantu, Costin Raiciu

Hyperscalers dominate the landscape of large network deployments, yet they rarely share data or insights about the challenges they face. In light of this supremacy, what problems can we find to solve in this space? We take an unconventional approach to find relevant research directions, starting from public plans to build a $100 billion datacenter for machine learning applications. Leveraging the language models scaling laws, we discover what workloads such a datacenter might carry and explore the challenges one may encounter in doing so, with a focus on networking research. We conclude that building the datacenter and training such models is technically possible, but this requires novel wide-area transports for inter-DC communication, a multipath transport and novel datacenter topologies for intra-datacenter communication, high speed scale-up networks and transports, outlining a rich research agenda for the networking community.

♻ ☆ MoC-System: Efficient Fault Tolerance for Sparse Mixture-of-Experts Model Training

Weilin Cai, Le Qin, Jiayi Huang

As large language models continue to scale up, distributed training systems have expanded beyond 10k nodes, intensifying the importance of fault tolerance. Checkpoint has emerged as the predominant fault tolerance strategy, with extensive studies dedicated to optimizing its efficiency. However, the advent of the sparse Mixture-of-Experts (MoE) model presents new challenges due to the substantial increase in model size, despite comparable computational demands to dense models. In this work, we propose the Mixture-of-Checkpoint System (MoC-System) to orchestrate the vast array of checkpoint shards produced in distributed training systems. MoC-System features a novel Partial Experts Checkpointing (PEC) mechanism, an algorithm-system co-design that strategically saves a selected subset of experts, effectively reducing the MoE checkpoint size to levels comparable with dense models. Incorporating hybrid parallel strategies, MoC-System involves fully sharded checkpointing strategies to evenly distribute the workload across distributed ranks. Furthermore, MoC-System introduces a two-level checkpointing management method that asynchronously handles in-memory snapshots and persistence processes. We build MoC-System upon the Megatron-DeepSpeed framework, achieving up to a 98.9% reduction in overhead for each checkpointing process compared to the original method, during MoE model training with ZeRO-2 data parallelism and expert parallelism. Additionally, extensive empirical analyses substantiate that our methods enhance efficiency while maintaining comparable model accuracy, even achieving an average accuracy increase of 1.08% on downstream tasks.

☆ POD-Attention: Unlocking Full Prefill-Decode Overlap for Faster LLM Inference

Aditya K Kamath, Ramya Prabhu, Jayashree Mohan, Simon Peter, Ramachandran Ramjee, Ashish Panwar

Each request in LLM inference goes through two phases: compute-bound prefill and memory-bandwidth-bound decode. To improve GPU utilization, recent systems use hybrid batching that combines the prefill and decode phases of different requests into the same batch. Hybrid batching works well for linear operations as it amortizes the cost of loading model weights from HBM. However, attention computation in hybrid batches remains inefficient because existing attention kernels are optimized for either prefill or decode. In this paper, we present POD-Attention -- the first GPU kernel that efficiently computes attention for hybrid batches. POD-Attention aims to maximize the utilization of both compute and memory bandwidth by carefully allocating the GPU's resources such that prefill and decode operations happen concurrently on the same multiprocessor. We integrate POD-Attention in a state-of-the-art LLM inference scheduler Sarathi-Serve. POD-Attention speeds up attention computation by up to 75% (mean 28%) and increases LLM serving throughput by up to 22% in offline inference. In online inference, POD-Attention enables lower time-to-first-token (TTFT), time-between-tokens (TBT), and request execution latency versus Sarathi-Serve.

☆ Optimal Fault-Tolerant Dispersion on Oriented Grids

Rik Banerjee, Manish Kumar, Anisur Rahaman Molla

Dispersion of mobile robots over the nodes of an anonymous graph is an important problem and turns out to be a crucial subroutine for designing efficient algorithms for many fundamental graph problems via mobile robots. In this problem, starting from an arbitrary initial distribution of $n$ robots across the $n$ nodes, the goal is to achieve a final configuration where each node holds at most one robot. This paper investigates the dispersion problem on an oriented grid, considering the possibility of robot failures (crashes) at any time during the algorithm's execution. We present a crash-tolerant dispersion algorithm that solves the dispersion problem on an anonymous oriented grid in $O(\sqrt{n})$ time and using $O(\log n)$ bits of memory per robot. The algorithm is optimal in terms of both time and memory per robot. We further extend this algorithm to deal with weak Byzantine robots. The weak Byzantine fault dispersion algorithm takes optimal $O(\sqrt{n})$ rounds but requires $O(n\log n)$ bits of memory per robot.

comment: arXiv admin note: substantial text overlap with arXiv:2405.02002

☆ Deoxys: A Causal Inference Engine for Unhealthy Node Mitigation in Large-scale Cloud Infrastructure

Chaoyun Zhang, Randolph Yao, Si Qin, Ze Li, Shekhar Agrawal, Binit R. Mishra, Tri Tran, Minghua Ma, Qingwei Lin, Murali Chintalapati, Dongmei Zhang

The presence of unhealthy nodes in cloud infrastructure signals the potential failure of machines, which can significantly impact the availability and reliability of cloud services, resulting in negative customer experiences. Effectively addressing unhealthy node mitigation is therefore vital for sustaining cloud system performance. This paper introduces Deoxys, a causal inference engine tailored to recommending mitigation actions for unhealthy node in cloud systems to minimize virtual machine downtime and interruptions during unhealthy events. It employs double machine learning combined with causal forest to produce precise and reliable mitigation recommendations based solely on limited observational data collected from the historical unhealthy events. To enhance the causal inference model, Deoxys further incorporates a policy fallback mechanism based on model uncertainty and action overriding mechanisms to (i) improve the reliability of the system, and (ii) strike a good tradeoff between downtime reduction and resource utilization, thereby enhancing the overall system performance. After deploying Deoxys in a large-scale cloud infrastructure at Microsoft, our observations demonstrate that Deoxys significantly reduces average VM downtime by 53% compared to a legacy policy, while leading to 49.5% lower VM interruption rate. This substantial improvement enhances the reliability and stability of cloud platforms, resulting in a seamless customer experience.

☆ Signature-based IaaS Performance Change Detection

Sheik Mohammad Mostakim Fattah, Athman Bouguettaya

We propose a novel change detection framework to identify changes in the long-term performance behavior of an IaaS service. An IaaS service's long-term performance behavior is represented by an IaaS performance signature. The proposed framework leverages time series similarity measures and a sliding window technique to detect changes in IaaS performance signatures. We introduce a new IaaS performance noise model that enables the proposed framework to distinguish between performance noise and actual changes in performance. The proposed framework utilizes a novel Signal-to-Noise Ratio (SNR) based approach to detect changes when prior knowledge about performance noise is available. A set of experiments is conducted using real-world datasets to demonstrate the effectiveness of the proposed change detection framework.

comment: accepted in ACM transaction on Internet Technology in October 2024. arXiv admin note: text overlap with arXiv:2007.11705

☆ Efficiently Scheduling Parallel DAG Tasks on Identical Multiprocessors

Shardul Lendve, Konstantinos Bletsas, Pedro F. Souto

Parallel real-time embedded applications can be modelled as directed acyclic graphs (DAGs) whose nodes model subtasks and whose edges model precedence constraints among subtasks. Efficiently scheduling such parallel tasks can be challenging in itself, particularly in hard real-time systems where it must be ensured offline that the deadlines of the parallel applications will be met at run time. In this paper, we tackle the problem of scheduling DAG tasks on identical multiprocessor systems efficiently, in terms of processor utilisation. We propose a new algorithm that attempts to use dedicated processor clusters for high-utilisation tasks, as in federated scheduling, but is also capable of reclaiming the processing capacity lost to fragmentation, by splitting the execution of parallel tasks over different existing clusters, in a manner inspired by semi-partitioned C=D scheduling (originally devised for non-parallel tasks). In the experiments with synthetic DAG task sets, our Segmented-Flattened-and-Split scheduling approach achieves a significantly higher scheduling success ratio than federated scheduling.

comment: Version submitted to RTNS 2024, on 16/08/2024 (with some typos fixed)

♻ ☆ TiMePReSt: Time and Memory Efficient Pipeline Parallel DNN Training with Removed Staleness

Ankita Dutta, Nabendu Chaki, Rajat K. De

DNN training is time-consuming and requires efficient multi-accelerator parallelization, where a single training iteration is split over available accelerators. Current approaches often parallelize training using intra-batch parallelization. Combining inter-batch and intra-batch pipeline parallelism is common to further improve training throughput. In this article, we develop a system, called TiMePReSt, that combines them in a novel way which helps to better overlap computation and communication, and limits the amount of communication. The traditional pipeline-parallel training of DNNs maintains similar working principle as sequential or conventional training of DNNs by maintaining consistent weight versions in forward and backward passes of a mini-batch. Thus, it suffers from high GPU memory footprint during training. In this paper, experimental study demonstrates that compromising weight consistency doesn't decrease prediction capability of a parallelly trained DNN. Moreover, TiMePReSt overcomes GPU memory overhead and achieves zero weight staleness. State-of-the-art techniques often become costly in terms of training time. In order to address this issue, TiMePReSt introduces a variant of intra-batch parallelism that parallelizes the forward pass of each mini-batch by decomposing it into smaller micro-batches. A novel synchronization method between forward and backward passes reduces training time in TiMePReSt. The occurrence of multiple sequence problem and its relation with version difference have been observed in TiMePReSt. This paper presents a mathematical relationship between the number of micro-batches and worker machines, highlighting the variation in version difference. A mathematical expression has been developed to calculate version differences for various combinations of these two without creating diagrams for all combinations.

♻ ☆ Harnessing Your DRAM and SSD for Sustainable and Accessible LLM Inference with Mixed-Precision and Multi-level Caching

Jie Peng, Zhang Cao, Huaizhi Qu, Zhengyu Zhang, Chang Guo, Yanyong Zhang, Zhichao Cao, Tianlong Chen

Although Large Language Models (LLMs) have demonstrated remarkable capabilities, their massive parameter counts and associated extensive computing make LLMs' deployment the main part of carbon emission from nowadays AI applications. Compared to modern GPUs like H$100$, it would be significantly carbon-sustainable if we could leverage old-fashioned GPUs such as M$40$ (as shown in Figure 1, M$40$ only has one third carbon emission of H$100$'s) for LLM servings. However, the limited High Bandwidth Memory (HBM) available on such GPU often cannot support the loading of LLMs due to the gigantic model size and intermediate activation data, making their serving challenging. For instance, a LLaMA2 model with $70$B parameters typically requires $128$GB for inference, which substantially surpasses $24$GB HBM in a $3090$ GPU and remains infeasible even considering the additional $64$GB DRAM. To address this challenge, this paper proposes a mixed-precision with a model modularization algorithm to enable LLM inference on outdated hardware with resource constraints. (The precision denotes the numerical precision like FP16, INT8, INT4) and multi-level caching (M2Cache).) Specifically, our M2Cache first modulizes neurons in LLM and creates their importance ranking. Then, it adopts a dynamic sparse mixed-precision quantization mechanism in weight space to reduce computational demands and communication overhead at each decoding step. It collectively lowers the operational carbon emissions associated with LLM inference. Moreover, M2Cache introduces a three-level cache management system with HBM, DRAM, and SSDs that complements the dynamic sparse mixed-precision inference. To enhance communication efficiency, M2Cache maintains a neuron-level mixed-precision LRU cache in HBM, a larger layer-aware cache in DRAM, and a full model in SSD.

comment: 24 pages, 13 figures

♻ ☆ Large Scale Multi-GPU Based Parallel Traffic Simulation for Accelerated Traffic Assignment and Propagation

Xuan Jiang, Raja Sengupta, James Demmel, Samuel Williams

Traffic propagation simulation is crucial for urban planning, enabling congestion analysis, travel time estimation, and route optimization. Traditional micro-simulation frameworks are limited to main roads due to the complexity of urban mobility and large-scale data. We introduce the Large Scale Multi-GPU Parallel Computing based Regional Scale Traffic Simulation Framework (LPSim), a scalable tool that leverages GPU parallel computing to simulate extensive traffic networks with high fidelity and reduced computation time. LPSim performs millions of vehicle dynamics simulations simultaneously, outperforming CPU-based methods. It can complete simulations of 2.82 million trips in 6.28 minutes using a single GPU, and 9.01 million trips in 21.16 minutes on dual GPUs. LPSim is also tested on dual NVIDIA A100 GPUs, achieving simulations about 113 times faster than traditional CPU methods. This demonstrates its scalability and efficiency for large-scale applications, making LPSim a valuable resource for researchers and planners. Code: https://github.com/Xuan-1998/LPSim

♻ ☆ Cold Start Latency in Serverless Computing: A Systematic Review, Taxonomy, and Future Directions

Muhammed Golec, Guneet Kaur Walia, Mohit Kumar, Felix Cuadrado, Sukhpal Singh Gill, Steve Uhlig

Recently, academics and the corporate sector have paid attention to serverless computing, which enables dynamic scalability and an economic model. In serverless computing, users only pay for the time they actually use resources, enabling zero scaling to optimise cost and resource utilisation. However, this approach also introduces the serverless cold start problem. Researchers have developed various solutions to address the cold start problem, yet it remains an unresolved research area. In this article, we propose a systematic literature review on clod start latency in serverless computing. Furthermore, we create a detailed taxonomy of approaches to cold start latency, which we use to investigate existing techniques for reducing the cold start time and frequency. We have classified the current studies on cold start latency into several categories such as caching and application-level optimisation-based solutions, as well as Artificial Intelligence (AI)/Machine Learning (ML)-based solutions. Moreover, we have analyzed the impact of cold start latency on quality of service, explored current cold start latency mitigation methods, datasets, and implementation platforms, and classified them into categories based on their common characteristics and features. Finally, we outline the open challenges and highlight the possible future directions.

comment: Preprint Version Accepted for Publication in ACM Computing Survey, 2024

♻ ☆ I've Got 99 Problems But FLOPS Ain't One

Alexandru M. Gherghescu, Vlad-Andrei Bădoiu, Alexandru Agache, Mihai-Valentin Dumitru, Iuliu Vasilescu, Radu Mantu, Costin Raiciu

Hyperscalers dominate the landscape of large network deployments, yet they rarely share data or insights about the challenges they face. In light of this supremacy, what problems can we find to solve in this space? We take an unconventional approach to find relevant research directions, starting from public plans to build a $100 billion datacenter for machine learning applications. Leveraging the language models scaling laws, we discover what workloads such a datacenter might carry and explore the challenges one may encounter in doing so, with a focus on networking research. We conclude that building the datacenter and training such models is technically possible, but this requires novel wide-area transports for inter-DC communication, a multipath transport and novel datacenter topologies for intra-datacenter communication, high speed scale-up networks and transports, outlining a rich research agenda for the networking community.

♻ ☆ MoC-System: Efficient Fault Tolerance for Sparse Mixture-of-Experts Model Training

Weilin Cai, Le Qin, Jiayi Huang

As large language models continue to scale up, distributed training systems have expanded beyond 10k nodes, intensifying the importance of fault tolerance. Checkpoint has emerged as the predominant fault tolerance strategy, with extensive studies dedicated to optimizing its efficiency. However, the advent of the sparse Mixture-of-Experts (MoE) model presents new challenges due to the substantial increase in model size, despite comparable computational demands to dense models. In this work, we propose the Mixture-of-Checkpoint System (MoC-System) to orchestrate the vast array of checkpoint shards produced in distributed training systems. MoC-System features a novel Partial Experts Checkpointing (PEC) mechanism, an algorithm-system co-design that strategically saves a selected subset of experts, effectively reducing the MoE checkpoint size to levels comparable with dense models. Incorporating hybrid parallel strategies, MoC-System involves fully sharded checkpointing strategies to evenly distribute the workload across distributed ranks. Furthermore, MoC-System introduces a two-level checkpointing management method that asynchronously handles in-memory snapshots and persistence processes. We build MoC-System upon the Megatron-DeepSpeed framework, achieving up to a 98.9% reduction in overhead for each checkpointing process compared to the original method, during MoE model training with ZeRO-2 data parallelism and expert parallelism. Additionally, extensive empirical analyses substantiate that our methods enhance efficiency while maintaining comparable model accuracy, even achieving an average accuracy increase of 1.08% on downstream tasks.

☆ Can Uncertainty Quantification Enable Better Learning-based Index Tuning?

Tao Yu, Zhaonian Zou, Hao Xiong

Index tuning is crucial for optimizing database performance by selecting optimal indexes based on workload. The key to this process lies in an accurate and efficient benefit estimator. Traditional methods relying on what-if tools often suffer from inefficiency and inaccuracy. In contrast, learning-based models provide a promising alternative but face challenges such as instability, lack of interpretability, and complex management. To overcome these limitations, we adopt a novel approach: quantifying the uncertainty in learning-based models' results, thereby combining the strengths of both traditional and learning-based methods for reliable index tuning. We propose Beauty, the first uncertainty-aware framework that enhances learning-based models with uncertainty quantification and uses what-if tools as a complementary mechanism to improve reliability and reduce management complexity. Specifically, we introduce a novel method that combines AutoEncoder and Monte Carlo Dropout to jointly quantify uncertainty, tailored to the characteristics of benefit estimation tasks. In experiments involving sixteen models, our approach outperformed existing uncertainty quantification methods in the majority of cases. We also conducted index tuning tests on six datasets. By applying the Beauty framework, we eliminated worst-case scenarios and more than tripled the occurrence of best-case scenarios.

comment: 14 pages, 11 figures

☆ Graphusion: A RAG Framework for Knowledge Graph Construction with a Global Perspective

Rui Yang, Boming Yang, Aosong Feng, Sixun Ouyang, Moritz Blum, Tianwei She, Yuang Jiang, Freddy Lecue, Jinghui Lu, Irene Li

Knowledge Graphs (KGs) are crucial in the field of artificial intelligence and are widely used in downstream tasks, such as question-answering (QA). The construction of KGs typically requires significant effort from domain experts. Large Language Models (LLMs) have recently been used for Knowledge Graph Construction (KGC). However, most existing approaches focus on a local perspective, extracting knowledge triplets from individual sentences or documents, missing a fusion process to combine the knowledge in a global KG. This work introduces Graphusion, a zero-shot KGC framework from free text. It contains three steps: in Step 1, we extract a list of seed entities using topic modeling to guide the final KG includes the most relevant entities; in Step 2, we conduct candidate triplet extraction using LLMs; in Step 3, we design the novel fusion module that provides a global view of the extracted knowledge, incorporating entity merging, conflict resolution, and novel triplet discovery. Results show that Graphusion achieves scores of 2.92 and 2.37 out of 3 for entity extraction and relation recognition, respectively. Moreover, we showcase how Graphusion could be applied to the Natural Language Processing (NLP) domain and validate it in an educational scenario. Specifically, we introduce TutorQA, a new expert-verified benchmark for QA, comprising six tasks and a total of 1,200 QA pairs. Using the Graphusion-constructed KG, we achieve a significant improvement on the benchmark, for example, a 9.2% accuracy improvement on sub-graph completion.

comment: arXiv admin note: substantial text overlap with arXiv:2407.10794

♻ ☆ CUBIT: Concurrent Updatable Bitmap Indexing (Extended Version)

Junchang Wang, Manos Athanassoulis

Bitmap indexes are widely used for read-intensive analytical workloads because they are clustered and offer efficient reads with a small memory footprint. However, they are notoriously inefficient to update. As analytical applications are increasingly fused with transactional applications, leading to the emergence of hybrid transactional/analytical processing (HTAP), it is desirable that bitmap indexes support efficient concurrent real-time updates. In this paper, we propose Concurrent Updatable Bitmap indexing (CUBIT) that offers efficient real-time updates that scale with the number of CPU cores used and do not interfere with queries. Our design relies on three principles. First, we employ a horizontal bitwise representation of updated bits, which enables efficient atomic updates without locking entire bitvectors. Second, we propose a lightweight snapshotting mechanism that allows queries (including range queries) to run on separate snapshots and provides a wait-free progress guarantee. Third, we consolidate updates in a latch-free manner, providing a strong progress guarantee. Our evaluation shows that CUBIT offers 3x - 16x higher throughput and 3x - 220x lower latency than state-of-the-art updatable bitmap indexes. CUBIT's update-friendly nature widens the applicability of bitmap indexing. Experimenting with OLAP workloads with standard, batched updates shows that CUBIT overcomes the maintenance downtime and outperforms DuckDB by 1.2x - 2.7x on TPC-H. For HTAP workloads with real-time updates, CUBIT achieves 2x - 11x performance improvement over the state-of-the-art approaches.

♻ ☆ Multiversion Hindsight Logging for Continuous Training

Rolando Garcia, Anusha Dandamudi, Gabriel Matute, Lehan Wan, Joseph Gonzalez, Joseph M. Hellerstein, Koushik Sen

Production Machine Learning involves continuous training: hosting multiple versions of models over time, often with many model versions running at once. When model performance does not meet expectations, Machine Learning Engineers (MLEs) debug issues by exploring and analyzing numerous prior versions of code and training data to identify root causes and mitigate problems. Traditional debugging and logging tools often fall short in managing this experimental, multi-version context. FlorDB introduces Multiversion Hindsight Logging, which allows engineers to use the most recent version's logging statements to query past versions, even when older versions logged different data. Log statement propagation enables consistent injection of logging statements into past code versions, regardless of changes to the codebase. Once log statements are propagated across code versions, the remaining challenge in Multiversion Hindsight Logging is to efficiently replay the new log statements based on checkpoints from previous runs. Finally, a coherent user experience is required to help MLEs debug across all versions of code and data. To this end, FlorDB presents a unified relational model for efficient handling of historical queries, offering a comprehensive view of the log history to simplify the exploration of past code iterations. We present a performance evaluation on diverse benchmarks confirming its scalability and the ability to deliver real-time query responses, leveraging query-based filtering and checkpoint-based parallelism for efficient replay.

☆ ALTA: Compiler-Based Analysis of Transformers

Peter Shaw, James Cohan, Jacob Eisenstein, Kenton Lee, Jonathan Berant, Kristina Toutanova

We propose a new programming language called ALTA and a compiler that can map ALTA programs to Transformer weights. ALTA is inspired by RASP, a language proposed by Weiss et al. (2021), and Tracr (Lindner et al., 2023), a compiler from RASP programs to Transformer weights. ALTA complements and extends this prior work, offering the ability to express loops and to compile programs to Universal Transformers, among other advantages. ALTA allows us to constructively show how Transformers can represent length-invariant algorithms for computing parity and addition, as well as a solution to the SCAN benchmark of compositional generalization tasks, without requiring intermediate scratchpad decoding steps. We also propose tools to analyze cases where the expressibility of an algorithm is established, but end-to-end training on a given training set fails to induce behavior consistent with the desired algorithm. To this end, we explore training from ALTA execution traces as a more fine-grained supervision signal. This enables additional experiments and theoretical analyses relating the learnability of various algorithms to data availability and modeling decisions, such as positional encodings. We make the ALTA framework -- language specification, symbolic interpreter, and weight compiler -- available to the community to enable further applications and insights.

☆ Leveraging Skills from Unlabeled Prior Data for Efficient Online Exploration

Max Wilcoxson, Qiyang Li, Kevin Frans, Sergey Levine

Unsupervised pretraining has been transformative in many supervised domains. However, applying such ideas to reinforcement learning (RL) presents a unique challenge in that fine-tuning does not involve mimicking task-specific data, but rather exploring and locating the solution through iterative self-improvement. In this work, we study how unlabeled prior trajectory data can be leveraged to learn efficient exploration strategies. While prior data can be used to pretrain a set of low-level skills, or as additional off-policy data for online RL, it has been unclear how to combine these ideas effectively for online exploration. Our method SUPE (Skills from Unlabeled Prior data for Exploration) demonstrates that a careful combination of these ideas compounds their benefits. Our method first extracts low-level skills using a variational autoencoder (VAE), and then pseudo-relabels unlabeled trajectories using an optimistic reward model, transforming prior data into high-level, task-relevant examples. Finally, SUPE uses these transformed examples as additional off-policy data for online RL to learn a high-level policy that composes pretrained low-level skills to explore efficiently. We empirically show that SUPE reliably outperforms prior strategies, successfully solving a suite of long-horizon, sparse-reward tasks. Code: https://github.com/rail-berkeley/supe.

comment: 23 pages, 10 figures

☆ TP-Eval: Tap Multimodal LLMs' Potential in Evaluation by Customizing Prompts

Yuxuan Xie, Tianhua Li, Wenqi Shao, Kaipeng Zhang

Recently, multimodal large language models (MLLMs) have received much attention for their impressive capabilities. The evaluation of MLLMs is becoming critical to analyzing attributes of MLLMs and providing valuable insights. However, current benchmarks overlook the problem of prompt sensitivity - minor prompt variations may lead to significant performance fluctuations. Thus, inappropriate prompts may obscure the models' capabilities, underestimating the models' performance. Moreover, different models have different preferences for different prompts, and thus, using the same prompt for all models will cause evaluation bias. This paper analyzes this deficiency in existing benchmarks and further introduces a new evaluation framework named TP-Eval, which introduces a prompt customization method to reduce evaluation biases and tap models' potential. TP-Eval will rewrite the original prompts to different customized prompts for different models. In particular, we propose some well-designed modules for prompt customization tailored to the scenario of MLLM evaluation. Extensive experiments demonstrate the effectiveness of our approach to uncovering models' capabilities, and TP-Eval should benefit the community in developing more comprehensive and convincing MLLM evaluation benchmarks.

☆ Training Free Guided Flow Matching with Optimal Control

Luran Wang, Chaoran Cheng, Yizhen Liao, Yanru Qu, Ge Liu

Controlled generation with pre-trained Diffusion and Flow Matching models has vast applications. One strategy for guiding ODE-based generative models is through optimizing a target loss $R(x_1)$ while staying close to the prior distribution. Along this line, some recent work showed the effectiveness of guiding flow model by differentiating through its ODE sampling process. Despite the superior performance, the theoretical understanding of this line of methods is still preliminary, leaving space for algorithm improvement. Moreover, existing methods predominately focus on Euclidean data manifold, and there is a compelling need for guided flow methods on complex geometries such as SO(3), which prevails in high-stake scientific applications like protein design. We present OC-Flow, a general and theoretically grounded training-free framework for guided flow matching using optimal control. Building upon advances in optimal control theory, we develop effective and practical algorithms for solving optimal control in guided ODE-based generation and provide a systematic theoretical analysis of the convergence guarantee in both Euclidean and SO(3). We show that existing backprop-through-ODE methods can be interpreted as special cases of Euclidean OC-Flow. OC-Flow achieved superior performance in extensive experiments on text-guided image manipulation, conditional molecule generation, and all-atom peptide design.

☆ Beyond position: how rotary embeddings shape representations and memory in autoregressive transfomers

Valeria Ruscio, Fabrizio Silvestri

Rotary Positional Embeddings (RoPE) enhance positional encoding in Transformer models, yet their full impact on model dynamics remains underexplored. This paper studies how RoPE introduces position-dependent rotations, causing phase shifts in token embeddings that influence higher-frequency components within the model's internal representations. Through spectral analysis, we demonstrate that RoPE's rotation matrices induce oscillatory behaviors in embeddings, affecting information retention across layers and shaping temporal modeling capabilities. We show that activation functions in feed-forward networks interact with RoPE-modulated embeddings to generate harmonics, leading to constructive or destructive interference based on phase alignment. Our findings reveal that phase alignment amplifies activations and sharpens attention, while misalignment weakens activations and disrupts focus on positional patterns. This study underscores the importance of frequency components as intrinsic elements of model behavior, offering new insights beyond traditional analyses.

☆ SPIRE: Synergistic Planning, Imitation, and Reinforcement Learning for Long-Horizon Manipulation

Zihan Zhou, Animesh Garg, Dieter Fox, Caelan Garrett, Ajay Mandlekar

Robot learning has proven to be a general and effective technique for programming manipulators. Imitation learning is able to teach robots solely from human demonstrations but is bottlenecked by the capabilities of the demonstrations. Reinforcement learning uses exploration to discover better behaviors; however, the space of possible improvements can be too large to start from scratch. And for both techniques, the learning difficulty increases proportional to the length of the manipulation task. Accounting for this, we propose SPIRE, a system that first uses Task and Motion Planning (TAMP) to decompose tasks into smaller learning subproblems and second combines imitation and reinforcement learning to maximize their strengths. We develop novel strategies to train learning agents when deployed in the context of a planning system. We evaluate SPIRE on a suite of long-horizon and contact-rich robot manipulation problems. We find that SPIRE outperforms prior approaches that integrate imitation learning, reinforcement learning, and planning by 35% to 50% in average task performance, is 6 times more data efficient in the number of human demonstrations needed to train proficient agents, and learns to complete tasks nearly twice as efficiently. View https://sites.google.com/view/spire-corl-2024 for more details.

comment: Conference on Robot Learning (CoRL) 2024

☆ Explaining Bayesian Networks in Natural Language using Factor Arguments. Evaluation in the medical domain

Jaime Sevilla, Nikolay Babakov, Ehud Reiter, Alberto Bugarin

In this paper, we propose a model for building natural language explanations for Bayesian Network Reasoning in terms of factor arguments, which are argumentation graphs of flowing evidence, relating the observed evidence to a target variable we want to learn about. We introduce the notion of factor argument independence to address the outstanding question of defining when arguments should be presented jointly or separately and present an algorithm that, starting from the evidence nodes and a target node, produces a list of all independent factor arguments ordered by their strength. Finally, we implemented a scheme to build natural language explanations of Bayesian Reasoning using this approach. Our proposal has been validated in the medical domain through a human-driven evaluation study where we compare the Bayesian Network Reasoning explanations obtained using factor arguments with an alternative explanation method. Evaluation results indicate that our proposed explanation approach is deemed by users as significantly more useful for understanding Bayesian Network Reasoning than another existing explanation method it is compared to.

comment: First Workshop on Explainable Artificial Intelligence for the medical domain - EXPLIMED. THE 27TH EUROPEAN CONFERENCE ON ARTIFICIAL INTELLIGENCE

☆ Key Algorithms for Keyphrase Generation: Instruction-Based LLMs for Russian Scientific Keyphrases

Anna Glazkova, Dmitry Morozov, Timur Garipov

Keyphrase selection is a challenging task in natural language processing that has a wide range of applications. Adapting existing supervised and unsupervised solutions for the Russian language faces several limitations due to the rich morphology of Russian and the limited number of training datasets available. Recent studies conducted on English texts show that large language models (LLMs) successfully address the task of generating keyphrases. LLMs allow achieving impressive results without task-specific fine-tuning, using text prompts instead. In this work, we access the performance of prompt-based methods for generating keyphrases for Russian scientific abstracts. First, we compare the performance of zero-shot and few-shot prompt-based methods, fine-tuned models, and unsupervised methods. Then we assess strategies for selecting keyphrase examples in a few-shot setting. We present the outcomes of human evaluation of the generated keyphrases and analyze the strengths and weaknesses of the models through expert assessment. Our results suggest that prompt-based methods can outperform common baselines even using simple text prompts.

comment: The 12th International Conference on Analysis of Images, Social Networks and Texts (AIST'2024)

☆ GraphTeam: Facilitating Large Language Model-based Graph Analysis via Multi-Agent Collaboration

Xin Li, Qizhi Chu, Yubin Chen, Yang Liu, Yaoqi Liu, Zekai Yu, Weize Chen, Chen Qian, Chuan Shi, Cheng Yang

Graphs are widely used for modeling relational data in real-world scenarios, such as social networks and urban computing. Existing LLM-based graph analysis approaches either integrate graph neural networks (GNNs) for specific machine learning tasks, limiting their transferability, or rely solely on LLMs' internal reasoning ability, resulting in suboptimal performance. To address these limitations, we take advantage of recent advances in LLM-based agents, which have shown capabilities of utilizing external knowledge or tools for problem solving. By simulating human problem-solving strategies such as analogy and collaboration, we propose a multi-agent system based on LLMs named GraphTeam, for graph analysis. GraphTeam consists of five LLM-based agents from three modules, and the agents with different specialities can collaborate with each other to address complex problems. Specifically, (1) input-output normalization module: the question agent extracts and refines four key arguments from the original question, facilitating the problem understanding, and the answer agent organizes the results to meet the output requirement; (2) external knowledge retrieval module: we first build a knowledge base consisting of relevant documentation and experience information, and then the search agent retrieves the most relevant entries for each question. (3) problem-solving module: given the retrieved information from search agent, the coding agent uses established algorithms via programming to generate solutions, and in case the coding agent does not work, the reasoning agent will directly compute the results without programming. Extensive experiments on six graph analysis benchmarks demonstrate that GraphTeam achieves state-of-the-art performance with an average 25.85% improvement over the best baseline in terms of accuracy. The code and data are available at https://github.com/BUPT-GAMMA/GraphTeam.

☆ Cross-lingual Transfer of Reward Models in Multilingual Alignment

Jiwoo Hong, Noah Lee, Rodrigo Martínez-Castaño, César Rodríguez, James Thorne

Reinforcement learning with human feedback (RLHF) is shown to largely benefit from precise reward models (RMs). However, recent studies in reward modeling schemes are skewed towards English, limiting the applicability of RLHF in multilingual alignments. In this work, we investigate the cross-lingual transfer of RMs trained in diverse languages, primarily from English. Our experimental results demonstrate the strong cross-lingual transfer of English RMs, exceeding target language RMs by 3~4% average increase in Multilingual RewardBench. Furthermore, we analyze the cross-lingual transfer of RMs through the representation shifts. Finally, we perform multilingual alignment to exemplify how cross-lingual transfer in RM propagates to enhanced multilingual instruction-following capability, along with extensive analyses on off-the-shelf RMs. We release the code, model, and data.

☆ Benchmarking Foundation Models on Exceptional Cases: Dataset Creation and Validation EMNLP 2024

Suho Kang, Jungyang Park, Joonseo Ha, SoMin Kim, JinHyeong Kim, Subeen Park, Kyungwoo Song

Foundation models (FMs) have achieved significant success across various tasks, leading to research on benchmarks for reasoning abilities. However, there is a lack of studies on FMs performance in exceptional scenarios, which we define as out-of-distribution (OOD) reasoning tasks. This paper is the first to address these cases, developing a novel dataset for evaluation of FMs across multiple modalities, including graphic novels, calligraphy, news articles, and lyrics. It includes tasks for instance classification, character recognition, token prediction, and text generation. The paper also proposes prompt engineering techniques like Chain-of-Thought (CoT) and CoT+Few-Shot to enhance performance. Validation of FMs using various methods revealed improvements. The code repository is accessible at: https://github.com/MLAI-Yonsei/ExceptionalBenchmark

comment: EMNLP 2024 Workshop Genbench(https://genbench.org/workshop_programme/)

☆ AI driven health recommender

K. Vignesh, B. Pranavi, Ch. Sreenidhi

As AI emerged as highest valued technology, We used that to create a web application that makes a patient work easier .It detects the disease name based on the symptoms given by the patient and recommends medication for respective disease, precautions to take, diet to follow and workouts to do, so the disease can be minimized. The web application is made with clean and Realtime data by using Machine learning as root. We used flask to create a user-friendly platform.

☆ Federated Transformer: Multi-Party Vertical Federated Learning on Practical Fuzzily Linked Data

Zhaomin Wu, Junyi Hou, Yiqun Diao, Bingsheng He

Federated Learning (FL) is an evolving paradigm that enables multiple parties to collaboratively train models without sharing raw data. Among its variants, Vertical Federated Learning (VFL) is particularly relevant in real-world, cross-organizational collaborations, where distinct features of a shared instance group are contributed by different parties. In these scenarios, parties are often linked using fuzzy identifiers, leading to a common practice termed as multi-party fuzzy VFL. Existing models generally address either multi-party VFL or fuzzy VFL between two parties. Extending these models to practical multi-party fuzzy VFL typically results in significant performance degradation and increased costs for maintaining privacy. To overcome these limitations, we introduce the Federated Transformer (FeT), a novel framework that supports multi-party VFL with fuzzy identifiers. FeT innovatively encodes these identifiers into data representations and employs a transformer architecture distributed across different parties, incorporating three new techniques to enhance performance. Furthermore, we have developed a multi-party privacy framework for VFL that integrates differential privacy with secure multi-party computation, effectively protecting local representations while minimizing associated utility costs. Our experiments demonstrate that the FeT surpasses the baseline models by up to 46\% in terms of accuracy when scaled to 50 parties. Additionally, in two-party fuzzy VFL settings, FeT also shows improved performance and privacy over cutting-edge VFL models.

☆ Dynamic Spectrum Access for Ambient Backscatter Communication-assisted D2D Systems with Quantum Reinforcement Learning

Nguyen Van Huynh, Bolun Zhang, Dinh-Hieu Tran, Dinh Thai Hoang, Diep N. Nguyen, Gan Zheng, Dusit Niyato, Quoc-Viet Pham

Spectrum access is an essential problem in device-to-device (D2D) communications. However, with the recent growth in the number of mobile devices, the wireless spectrum is becoming scarce, resulting in low spectral efficiency for D2D communications. To address this problem, this paper aims to integrate the ambient backscatter communication technology into D2D devices to allow them to backscatter ambient RF signals to transmit their data when the shared spectrum is occupied by mobile users. To obtain the optimal spectrum access policy, i.e., stay idle or access the shared spectrum and perform active transmissions or backscattering ambient RF signals for transmissions, to maximize the average throughput for D2D users, deep reinforcement learning (DRL) can be adopted. However, DRL-based solutions may require long training time due to the curse of dimensionality issue as well as complex deep neural network architectures. For that, we develop a novel quantum reinforcement learning (RL) algorithm that can achieve a faster convergence rate with fewer training parameters compared to DRL thanks to the quantum superposition and quantum entanglement principles. Specifically, instead of using conventional deep neural networks, the proposed quantum RL algorithm uses a parametrized quantum circuit to approximate an optimal policy. Extensive simulations then demonstrate that the proposed solution not only can significantly improve the average throughput of D2D devices when the shared spectrum is busy but also can achieve much better performance in terms of convergence rate and learning complexity compared to existing DRL-based methods.

comment: 12 pages, 7 figures

☆ Closed-form merging of parameter-efficient modules for Federated Continual Learning

Riccardo Salami, Pietro Buzzega, Matteo Mosconi, Jacopo Bonato, Luigi Sabetta, Simone Calderara

Model merging has emerged as a crucial technique in Deep Learning, enabling the integration of multiple models into a unified system while preserving performance and scalability. In this respect, the compositional properties of low-rank adaptation techniques (e.g., LoRA) have proven beneficial, as simple averaging LoRA modules yields a single model that mostly integrates the capabilities of all individual modules. Building on LoRA, we take a step further by imposing that the merged model matches the responses of all learned modules. Solving this objective in closed form yields an indeterminate system with A and B as unknown variables, indicating the existence of infinitely many closed-form solutions. To address this challenge, we introduce LoRM, an alternating optimization strategy that trains one LoRA matrix at a time. This allows solving for each unknown variable individually, thus finding a unique solution. We apply our proposed methodology to Federated Class-Incremental Learning (FCIL), ensuring alignment of model responses both between clients and across tasks. Our method demonstrates state-of-the-art performance across a range of FCIL scenarios.

☆ MCUBERT: Memory-Efficient BERT Inference on Commodity Microcontrollers

Zebin Yang, Renze Chen, Taiqiang Wu, Ngai Wong, Yun Liang, Runsheng Wang, Ru Huang, Meng Li

In this paper, we propose MCUBERT to enable language models like BERT on tiny microcontroller units (MCUs) through network and scheduling co-optimization. We observe the embedding table contributes to the major storage bottleneck for tiny BERT models. Hence, at the network level, we propose an MCU-aware two-stage neural architecture search algorithm based on clustered low-rank approximation for embedding compression. To reduce the inference memory requirements, we further propose a novel fine-grained MCU-friendly scheduling strategy. Through careful computation tiling and re-ordering as well as kernel design, we drastically increase the input sequence lengths supported on MCUs without any latency or accuracy penalty. MCUBERT reduces the parameter size of BERT-tiny and BERT-mini by 5.7$\times$ and 3.0$\times$ and the execution memory by 3.5$\times$ and 4.3$\times$, respectively. MCUBERT also achieves 1.5$\times$ latency reduction. For the first time, MCUBERT enables lightweight BERT models on commodity MCUs and processing more than 512 tokens with less than 256KB of memory.

comment: ICCAD 2024

☆ ExpertFlow: Optimized Expert Activation and Token Allocation for Efficient Mixture-of-Experts Inference

Xin He, Shunkang Zhang, Yuxin Wang, Haiyan Yin, Zihao Zeng, Shaohuai Shi, Zhenheng Tang, Xiaowen Chu, Ivor Tsang, Ong Yew Soon

Sparse Mixture of Experts (MoE) models, while outperforming dense Large Language Models (LLMs) in terms of performance, face significant deployment challenges during inference due to their high memory demands. Existing offloading techniques, which involve swapping activated and idle experts between the GPU and CPU, often suffer from rigid expert caching mechanisms. These mechanisms fail to adapt to dynamic routing, leading to inefficient cache utilization, or incur prohibitive costs for prediction training. To tackle these inference-specific challenges, we introduce ExpertFlow, a comprehensive system specifically designed to enhance inference efficiency by accommodating flexible routing and enabling efficient expert scheduling between CPU and GPU. This reduces overhead and boosts system performance. Central to our approach is a predictive routing path-based offloading mechanism that utilizes a lightweight predictor to accurately forecast routing paths before computation begins. This proactive strategy allows for real-time error correction in expert caching, significantly increasing cache hit ratios and reducing the frequency of expert transfers, thereby minimizing I/O overhead. Additionally, we implement a dynamic token scheduling strategy that optimizes MoE inference by rearranging input tokens across different batches. This method not only reduces the number of activated experts per batch but also improves computational efficiency. Our extensive experiments demonstrate that ExpertFlow achieves up to 93.72\% GPU memory savings and enhances inference speed by 2 to 10 times compared to baseline methods, highlighting its effectiveness and utility as a robust solution for resource-constrained inference scenarios.

comment: Mixture-of-Experts, Inference, Offloading

☆ SimRAG: Self-Improving Retrieval-Augmented Generation for Adapting Large Language Models to Specialized Domains

Ran Xu, Hui Liu, Sreyashi Nag, Zhenwei Dai, Yaochen Xie, Xianfeng Tang, Chen Luo, Yang Li, Joyce C. Ho, Carl Yang, Qi He

Retrieval-augmented generation (RAG) enhances the question-answering (QA) abilities of large language models (LLMs) by integrating external knowledge. However, adapting general-purpose RAG systems to specialized fields such as science and medicine poses unique challenges due to distribution shifts and limited access to domain-specific data. To tackle this, we propose SimRAG, a self-training approach that equips the LLM with joint capabilities of question answering and question generation for domain adaptation. Our method first fine-tunes the LLM on instruction-following, question-answering, and search-related data. Then, it prompts the same LLM to generate diverse domain-relevant questions from unlabeled corpora, with an additional filtering strategy to retain high-quality synthetic examples. By leveraging these synthetic examples, the LLM can improve their performance on domain-specific RAG tasks. Experiments on 11 datasets, spanning two backbone sizes and three domains, demonstrate that SimRAG outperforms baselines by 1.2\%--8.6\%.

comment: Work in Progress

☆ Benchmarking Floworks against OpenAI & Anthropic: A Novel Framework for Enhanced LLM Function Calling

Nirav Bhan, Shival Gupta, Sai Manaswini, Ritik Baba, Narun Yadav, Hillori Desai, Yash Choudhary, Aman Pawar, Sarthak Shrivastava, Sudipta Biswas

Large Language Models (LLMs) have shown remarkable capabilities in various domains, yet their economic impact has been limited by challenges in tool use and function calling. This paper introduces ThorV2, a novel architecture that significantly enhances LLMs' function calling abilities. We develop a comprehensive benchmark focused on HubSpot CRM operations to evaluate ThorV2 against leading models from OpenAI and Anthropic. Our results demonstrate that ThorV2 outperforms existing models in accuracy, reliability, latency, and cost efficiency for both single and multi-API calling tasks. We also show that ThorV2 is far more reliable and scales better to multistep tasks compared to traditional models. Our work offers the tantalizing possibility of more accurate function-calling compared to today's best-performing models using significantly smaller LLMs. These advancements have significant implications for the development of more capable AI assistants and the broader application of LLMs in real-world scenarios.

comment: 15 pages for main paper, 21 pages in total including references and appendix, 10 figures

☆ Optimizing Travel Itineraries with AI Algorithms in a Microservices Architecture: Balancing Cost, Time, Preferences, and Sustainability

Biman Barua, M. Shamim Kaiser

The objective of this research is how an implementation of AI algorithms in the microservices architecture enhances travel itineraries by cost, time, user preferences, and environmental sustainability. It uses machine learning models for both cost forecasting and personalization, genetic algorithm for optimization of the itinerary, and heuristics for sustainability checking. Primary evaluated parameters consist of latency, ability to satisfy user preferences, cost and environmental concern. The experimental results demonstrate an average of 4.5 seconds of response time on 1000 concurrent users and 92% of user preferences accuracy. The cost efficiency is proved, with 95% of provided trips being within the limits of the budget declared by the user. The system also implements some measures to alleviate negative externalities related to travel and 60% of offered travel plans had green options incorporated, resulting in the average 15% lower carbon emissions than the traditional travel plans offered. The genetic algorithm with time complexity O(g.p.f) provides the optimal solution in 100 generations. Every iteration improves the quality of the solution by 5%, thus enabling its effective use in optimization problems where time is measured in seconds. Finally, the system is designed to be fault-tolerant with functional 99.9% availability which allows the provision of services even when requirements are exceeded. Travel optimization platform is turned dynamic and efficient by this microservices based architecture which provides enhanced scaling, allows asynchronous communication and real time changes. Because of the incorporation of Ai, cost control and eco-friendliness approaches, the system addresses the different user needs in the present days travel business.

comment: 18 pages, 6 figures

☆ Multi-Continental Healthcare Modelling Using Blockchain-Enabled Federated Learning

Rui Sun, Zhipeng Wang, Hengrui Zhang, Ming Jiang, Yizhe Wen, Jiqun Zhang, Jiahao Sun, Shuoying Zhang, Erwu Liu, Kezhi Li

One of the biggest challenges of building artificial intelligence (AI) model in healthcare area is the data sharing. Since healthcare data is private, sensitive, and heterogeneous, collecting sufficient data for modelling is exhausted, costly, and sometimes impossible. In this paper, we propose a framework for global healthcare modelling using datasets from multi-continents (Europe, North America and Asia) while without sharing the local datasets, and choose glucose management as a study model to verify its effectiveness. Technically, blockchain-enabled federated learning is implemented with adaption to make it meet with the privacy and safety requirements of healthcare data, meanwhile rewards honest participation and penalize malicious activities using its on-chain incentive mechanism. Experimental results show that the proposed framework is effective, efficient, and privacy preserved. Its prediction accuracy is much better than the models trained from limited personal data and is similar to, and even slightly better than, the results from a centralized dataset. This work paves the way for international collaborations on healthcare projects, where additional data is crucial for reducing bias and providing benefits to humanity.

comment: Accepted by IEEE Global Blockchain Conference

☆ Guide for Defense (G4D): Dynamic Guidance for Robust and Balanced Defense in Large Language Models

He Cao, Weidi Luo, Yu Wang, Zijing Liu, Bing Feng, Yuan Yao, Yu Li

With the extensive deployment of Large Language Models (LLMs), ensuring their safety has become increasingly critical. However, existing defense methods often struggle with two key issues: (i) inadequate defense capabilities, particularly in domain-specific scenarios like chemistry, where a lack of specialized knowledge can lead to the generation of harmful responses to malicious queries. (ii) over-defensiveness, which compromises the general utility and responsiveness of LLMs. To mitigate these issues, we introduce a multi-agents-based defense framework, Guide for Defense (G4D), which leverages accurate external information to provide an unbiased summary of user intentions and analytically grounded safety response guidance. Extensive experiments on popular jailbreak attacks and benign datasets show that our G4D can enhance LLM's robustness against jailbreak attacks on general and domain-specific scenarios without compromising the model's general functionality.

☆ Addressing Asynchronicity in Clinical Multimodal Fusion via Individualized Chest X-ray Generation NeurIPS-24

Wenfang Yao, Chen Liu, Kejing Yin, William K. Cheung, Jing Qin

Integrating multi-modal clinical data, such as electronic health records (EHR) and chest X-ray images (CXR), is particularly beneficial for clinical prediction tasks. However, in a temporal setting, multi-modal data are often inherently asynchronous. EHR can be continuously collected but CXR is generally taken with a much longer interval due to its high cost and radiation dose. When clinical prediction is needed, the last available CXR image might have been outdated, leading to suboptimal predictions. To address this challenge, we propose DDL-CXR, a method that dynamically generates an up-to-date latent representation of the individualized CXR images. Our approach leverages latent diffusion models for patient-specific generation strategically conditioned on a previous CXR image and EHR time series, providing information regarding anatomical structures and disease progressions, respectively. In this way, the interaction across modalities could be better captured by the latent CXR generation process, ultimately improving the prediction performance. Experiments using MIMIC datasets show that the proposed model could effectively address asynchronicity in multimodal fusion and consistently outperform existing methods.

comment: Accepted by NeurIPS-24

☆ Leveraging Deep Learning for Time Series Extrinsic Regression in predicting photometric metallicity of Fundamental-mode RR Lyrae Stars

Lorenzo Monti, Tatiana Muraveva, Gisella Clementini, Alessia Garofalo

Astronomy is entering an unprecedented era of Big Data science, driven by missions like the ESA's Gaia telescope, which aims to map the Milky Way in three dimensions. Gaia's vast dataset presents a monumental challenge for traditional analysis methods. The sheer scale of this data exceeds the capabilities of manual exploration, necessitating the utilization of advanced computational techniques. In response to this challenge, we developed a novel approach leveraging deep learning to estimate the metallicity of fundamental mode (ab-type) RR Lyrae stars from their light curves in the Gaia optical G-band. Our study explores applying deep learning techniques, particularly advanced neural network architectures, in predicting photometric metallicity from time-series data. Our deep learning models demonstrated notable predictive performance, with a low mean absolute error (MAE) of 0.0565, the root mean square error (RMSE) achieved is 0.0765 and a high $R^2$ regression performance of 0.9401 measured by cross-validation. The weighted mean absolute error (wMAE) is 0.0563, while the weighted root mean square error (wRMSE) is 0.0763. These results showcase the effectiveness of our approach in accurately estimating metallicity values. Our work underscores the importance of deep learning in astronomical research, particularly with large datasets from missions like Gaia. By harnessing the power of deep learning methods, we can provide precision in analyzing vast datasets, contributing to more precise and comprehensive insights into complex astronomical phenomena.

comment: Sensors 2024, 24(16), 5203; (23 pages)

☆ Reinforcement Learning under Latent Dynamics: Toward Statistical and Algorithmic Modularity

Philip Amortila, Dylan J. Foster, Nan Jiang, Akshay Krishnamurthy, Zakaria Mhammedi

Real-world applications of reinforcement learning often involve environments where agents operate on complex, high-dimensional observations, but the underlying (''latent'') dynamics are comparatively simple. However, outside of restrictive settings such as small latent spaces, the fundamental statistical requirements and algorithmic principles for reinforcement learning under latent dynamics are poorly understood. This paper addresses the question of reinforcement learning under $\textit{general}$ latent dynamics from a statistical and algorithmic perspective. On the statistical side, our main negative result shows that most well-studied settings for reinforcement learning with function approximation become intractable when composed with rich observations; we complement this with a positive result, identifying latent pushforward coverability as a general condition that enables statistical tractability. Algorithmically, we develop provably efficient observable-to-latent reductions -- that is, reductions that transform an arbitrary algorithm for the latent MDP into an algorithm that can operate on rich observations -- in two settings: one where the agent has access to hindsight observations of the latent dynamics [LADZ23], and one where the agent can estimate self-predictive latent models [SAGHCB20]. Together, our results serve as a first step toward a unified statistical and algorithmic theory for reinforcement learning under latent dynamics.

☆ R-CoT: Reverse Chain-of-Thought Problem Generation for Geometric Reasoning in Large Multimodal Models

Linger Deng, Yuliang Liu, Bohan Li, Dongliang Luo, Liang Wu, Chengquan Zhang, Pengyuan Lyu, Ziyang Zhang, Gang Zhang, Errui Ding, Yingying Zhu, Xiang Bai

Existing Large Multimodal Models (LMMs) struggle with mathematical geometric reasoning due to a lack of high-quality image-text paired data. Current geometric data generation approaches, which apply preset templates to generate geometric data or use Large Language Models (LLMs) to rephrase questions and answers (Q&A), unavoidably limit data accuracy and diversity. To synthesize higher-quality data, we propose a two-stage Reverse Chain-of-Thought (R-CoT) geometry problem generation pipeline. First, we introduce GeoChain to produce high-fidelity geometric images and corresponding descriptions highlighting relations among geometric elements. We then design a Reverse A&Q method that reasons step-by-step based on the descriptions and generates questions in reverse from the reasoning results. Experiments demonstrate that the proposed method brings significant and consistent improvements on multiple LMM baselines, achieving new performance records in the 2B, 7B, and 8B settings. Notably, R-CoT-8B significantly outperforms previous state-of-the-art open-source mathematical models by 16.6% on MathVista and 9.2% on GeoQA, while also surpassing the closed-source model GPT-4o by an average of 13% across both datasets. The code is available at https://github.com/dle666/R-CoT.

☆ Lightweight Neural App Control

Filippos Christianos, Georgios Papoudakis, Thomas Coste, Jianye Hao, Jun Wang, Kun Shao

This paper introduces a novel mobile phone control architecture, termed ``app agents", for efficient interactions and controls across various Android apps. The proposed Lightweight Multi-modal App Control (LiMAC) takes as input a textual goal and a sequence of past mobile observations, such as screenshots and corresponding UI trees, to generate precise actions. To address the computational constraints inherent to smartphones, within LiMAC, we introduce a small Action Transformer (AcT) integrated with a fine-tuned vision-language model (VLM) for real-time decision-making and task execution. We evaluate LiMAC on two open-source mobile control datasets, demonstrating the superior performance of our small-form-factor approach against fine-tuned versions of open-source VLMs, such as Florence2 and Qwen2-VL. It also significantly outperforms prompt engineering baselines utilising closed-source foundation models like GPT-4o. More specifically, LiMAC increases the overall action accuracy by up to 19% compared to fine-tuned VLMs, and up to 42% compared to prompt-engineering baselines.

☆ Understanding Layer Significance in LLM Alignment

Guangyuan Shi, Zexin Lu, Xiaoyu Dong, Wenlong Zhang, Xuanyu Zhang, Yujie Feng, Xiao-Ming Wu

Aligning large language models (LLMs) through fine-tuning is essential for tailoring them to specific applications. Therefore, understanding what LLMs learn during the alignment process is crucial. Recent studies suggest that alignment primarily adjusts a model's presentation style rather than its foundational knowledge, indicating that only certain components of the model are significantly impacted. To delve deeper into LLM alignment, we propose to identify which layers within LLMs are most critical to the alignment process, thereby uncovering how alignment influences model behavior at a granular level. We propose a novel approach to identify the important layers for LLM alignment (ILA). It involves learning a binary mask for each incremental weight matrix in the LoRA algorithm, indicating the significance of each layer. ILA consistently identifies important layers across various alignment datasets, with nearly 90% overlap even with substantial dataset differences, highlighting fundamental patterns in LLM alignment. Experimental results indicate that freezing non-essential layers improves overall model performance, while selectively tuning the most critical layers significantly enhances fine-tuning efficiency with minimal performance loss.

☆ DataTales: A Benchmark for Real-World Intelligent Data Narration

Yajing Yang, Qian Liu, Min-Yen Kan

We introduce DataTales, a novel benchmark designed to assess the proficiency of language models in data narration, a task crucial for transforming complex tabular data into accessible narratives. Existing benchmarks often fall short in capturing the requisite analytical complexity for practical applications. DataTales addresses this gap by offering 4.9k financial reports paired with corresponding market data, showcasing the demand for models to create clear narratives and analyze large datasets while understanding specialized terminology in the field. Our findings highlights the significant challenge that language models face in achieving the necessary precision and analytical depth for proficient data narration, suggesting promising avenues for future model development and evaluation methodologies.

☆ ROCKET-1: Master Open-World Interaction with Visual-Temporal Context Prompting

Shaofei Cai, Zihao Wang, Kewei Lian, Zhancun Mu, Xiaojian Ma, Anji Liu, Yitao Liang

Vision-language models (VLMs) have excelled in multimodal tasks, but adapting them to embodied decision-making in open-world environments presents challenges. A key issue is the difficulty in smoothly connecting individual entities in low-level observations with abstract concepts required for planning. A common approach to address this problem is through the use of hierarchical agents, where VLMs serve as high-level reasoners that break down tasks into executable sub-tasks, typically specified using language and imagined observations. However, language often fails to effectively convey spatial information, while generating future images with sufficient accuracy remains challenging. To address these limitations, we propose visual-temporal context prompting, a novel communication protocol between VLMs and policy models. This protocol leverages object segmentation from both past and present observations to guide policy-environment interactions. Using this approach, we train ROCKET-1, a low-level policy that predicts actions based on concatenated visual observations and segmentation masks, with real-time object tracking provided by SAM-2. Our method unlocks the full potential of VLMs visual-language reasoning abilities, enabling them to solve complex creative tasks, especially those heavily reliant on spatial understanding. Experiments in Minecraft demonstrate that our approach allows agents to accomplish previously unattainable tasks, highlighting the effectiveness of visual-temporal context prompting in embodied decision-making. Codes and demos will be available on the project page: https://craftjarvis.github.io/ROCKET-1.

☆ TAGE: Trustworthy Attribute Group Editing for Stable Few-shot Image Generation

Ruicheng Zhang, Guoheng Huang, Yejing Huo, Xiaochen Yuan, Zhizhen Zhou, Xuhang Chen, Guo Zhong

Generative Adversarial Networks (GANs) have emerged as a prominent research focus for image editing tasks, leveraging the powerful image generation capabilities of the GAN framework to produce remarkable results.However, prevailing approaches are contingent upon extensive training datasets and explicit supervision, presenting a significant challenge in manipulating the diverse attributes of new image classes with limited sample availability. To surmount this hurdle, we introduce TAGE, an innovative image generation network comprising three integral modules: the Codebook Learning Module (CLM), the Code Prediction Module (CPM) and the Prompt-driven Semantic Module (PSM). The CPM module delves into the semantic dimensions of category-agnostic attributes, encapsulating them within a discrete codebook. This module is predicated on the concept that images are assemblages of attributes, and thus, by editing these category-independent attributes, it is theoretically possible to generate images from unseen categories. Subsequently, the CPM module facilitates naturalistic image editing by predicting indices of category-independent attribute vectors within the codebook. Additionally, the PSM module generates semantic cues that are seamlessly integrated into the Transformer architecture of the CPM, enhancing the model's comprehension of the targeted attributes for editing. With these semantic cues, the model can generate images that accentuate desired attributes more prominently while maintaining the integrity of the original category, even with a limited number of samples. We have conducted extensive experiments utilizing the Animal Faces, Flowers, and VGGFaces datasets. The results of these experiments demonstrate that our proposed method not only achieves superior performance but also exhibits a high degree of stability when compared to other few-shot image generation techniques.

comment: Accepted by International Conference on Signal Processing Systems Conference

☆ The Probabilistic Tsetlin Machine: A Novel Approach to Uncertainty Quantification

K. Darshana Abeyrathna, Sara El Mekkaoui, Andreas Hafver, Christian Agrell

Tsetlin Machines (TMs) have emerged as a compelling alternative to conventional deep learning methods, offering notable advantages such as smaller memory footprint, faster inference, fault-tolerant properties, and interpretability. Although various adaptations of TMs have expanded their applicability across diverse domains, a fundamental gap remains in understanding how TMs quantify uncertainty in their predictions. In response, this paper introduces the Probabilistic Tsetlin Machine (PTM) framework, aimed at providing a robust, reliable, and interpretable approach for uncertainty quantification. Unlike the original TM, the PTM learns the probability of staying on each state of each Tsetlin Automaton (TA) across all clauses. These probabilities are updated using the feedback tables that are part of the TM framework: Type I and Type II feedback. During inference, TAs decide their actions by sampling states based on learned probability distributions, akin to Bayesian neural networks when generating weight values. In our experimental analysis, we first illustrate the spread of the probabilities across TA states for the noisy-XOR dataset. Then we evaluate the PTM alongside benchmark models using both simulated and real-world datasets. The experiments on the simulated dataset reveal the PTM's effectiveness in uncertainty quantification, particularly in delineating decision boundaries and identifying regions of high uncertainty. Moreover, when applied to multiclass classification tasks using the Iris dataset, the PTM demonstrates competitive performance in terms of predictive entropy and expected calibration error, showcasing its potential as a reliable tool for uncertainty estimation. Our findings underscore the importance of selecting appropriate models for accurate uncertainty quantification in predictive tasks, with the PTM offering a particularly interpretable and effective solution.

comment: 12 pages, 5 figures, 6 tables, accepted and presented at ICAAI 2024, London

☆ RE-tune: Incremental Fine Tuning of Biomedical Vision-Language Models for Multi-label Chest X-ray Classification NeurIPS

Marco Mistretta, Andrew D. Bagdanov

In this paper we introduce RE-tune, a novel approach for fine-tuning pre-trained Multimodal Biomedical Vision-Language models (VLMs) in Incremental Learning scenarios for multi-label chest disease diagnosis. RE-tune freezes the backbones and only trains simple adaptors on top of the Image and Text encoders of the VLM. By engineering positive and negative text prompts for diseases, we leverage the ability of Large Language Models to steer the training trajectory. We evaluate RE-tune in three realistic incremental learning scenarios: class-incremental, label-incremental, and data-incremental. Our results demonstrate that Biomedical VLMs are natural continual learners and prevent catastrophic forgetting. RE-tune not only achieves accurate multi-label classification results, but also prioritizes patient privacy and it distinguishes itself through exceptional computational efficiency, rendering it highly suitable for broad adoption in real-world healthcare settings.

comment: Accepted for publication at Medical Imaging meets NeurIPS (NeurIPS23)

☆ PGDiffSeg: Prior-Guided Denoising Diffusion Model with Parameter-Shared Attention for Breast Cancer Segmentation

Feiyan Feng, Tianyu Liu, Hong Wang, Jun Zhao, Wei Li, Yanshen Sun

Early detection through imaging and accurate diagnosis is crucial in mitigating the high mortality rate associated with breast cancer. However, locating tumors from low-resolution and high-noise medical images is extremely challenging. Therefore, this paper proposes a novel PGDiffSeg (Prior-Guided Diffusion Denoising Model with Parameter-Shared Attention) that applies diffusion denoising methods to breast cancer medical image segmentation, accurately recovering the affected areas from Gaussian noise. Firstly, we design a parallel pipeline for noise processing and semantic information processing and propose a parameter-shared attention module (PSA) in multi-layer that seamlessly integrates these two pipelines. This integration empowers PGDiffSeg to incorporate semantic details at multiple levels during the denoising process, producing highly accurate segmentation maps. Secondly, we introduce a guided strategy that leverages prior knowledge to simulate the decision-making process of medical professionals, thereby enhancing the model's ability to locate tumor positions precisely. Finally, we provide the first-ever discussion on the interpretability of the generative diffusion model in the context of breast cancer segmentation. Extensive experiments have demonstrated the superiority of our model over the current state-of-the-art approaches, confirming its effectiveness as a flexible diffusion denoising method suitable for medical image research. Our code will be publicly available later.

☆ OmniFlatten: An End-to-end GPT Model for Seamless Voice Conversation

Qinglin Zhang, Luyao Cheng, Chong Deng, Qian Chen, Wen Wang, Siqi Zheng, Jiaqing Liu, Hai Yu, Chaohong Tan

Full-duplex spoken dialogue systems significantly advance over traditional turn-based dialogue systems, as they allow simultaneous bidirectional communication, closely mirroring human-human interactions. However, achieving low latency and natural interactions in full-duplex dialogue systems remains a significant challenge, especially considering human conversation dynamics such as interruptions, backchannels, and overlapping speech. In this paper, we introduce a novel End-to-End GPT-based model OmniFlatten for full-duplex conversation, capable of effectively modeling the complex behaviors inherent to natural conversations with low latency. To achieve full-duplex communication capabilities, we propose a multi-stage post-training scheme that progressively adapts a text-based large language model (LLM) backbone into a speech-text dialogue LLM, capable of generating text and speech in real time, without modifying the architecture of the backbone LLM. The training process comprises three stages: modality alignment, half-duplex dialogue learning, and full-duplex dialogue learning. Throughout all training stages, we standardize the data using a flattening operation, which allows us to unify the training methods and the model architecture across different modalities and tasks. Our approach offers a straightforward modeling technique and a promising research direction for developing efficient and natural end-to-end full-duplex spoken dialogue systems. Audio samples of dialogues generated by OmniFlatten can be found at this web site (https://omniflatten.github.io/).

comment: Work in progress

☆ Enhancing Federated Learning Convergence with Dynamic Data Queue and Data Entropy-driven Participant Selection

Charuka Herath, Xiaolan Liu, Sangarapillai Lambotharan, Yogachandran Rahulamathavan

Federated Learning (FL) is a decentralized approach for collaborative model training on edge devices. This distributed method of model training offers advantages in privacy, security, regulatory compliance, and cost-efficiency. Our emphasis in this research lies in addressing statistical complexity in FL, especially when the data stored locally across devices is not identically and independently distributed (non-IID). We have observed an accuracy reduction of up to approximately 10\% to 30\%, particularly in skewed scenarios where each edge device trains with only 1 class of data. This reduction is attributed to weight divergence, quantified using the Euclidean distance between device-level class distributions and the population distribution, resulting in a bias term (\(\delta_k\)). As a solution, we present a method to improve convergence in FL by creating a global subset of data on the server and dynamically distributing it across devices using a Dynamic Data queue-driven Federated Learning (DDFL). Next, we leverage Data Entropy metrics to observe the process during each training round and enable reasonable device selection for aggregation. Furthermore, we provide a convergence analysis of our proposed DDFL to justify their viability in practical FL scenarios, aiming for better device selection, a non-sub-optimal global model, and faster convergence. We observe that our approach results in a substantial accuracy boost of approximately 5\% for the MNIST dataset, around 18\% for CIFAR-10, and 20\% for CIFAR-100 with a 10\% global subset of data, outperforming the state-of-the-art (SOTA) aggregation algorithms.

comment: The Journal is submitted to IEEE Transactions in the Internet of Things

☆ Large Language Models Engineer Too Many Simple Features For Tabular Data

Jaris Küken, Lennart Purucker, Frank Hutter

Tabular machine learning problems often require time-consuming and labor-intensive feature engineering. Recent efforts have focused on using large language models (LLMs) to capitalize on their potential domain knowledge. At the same time, researchers have observed ethically concerning negative biases in other LLM-related use cases, such as text generation. These developments motivated us to investigate whether LLMs exhibit a bias that negatively impacts the performance of feature engineering. While not ethically concerning, such a bias could hinder practitioners from fully utilizing LLMs for automated data science. Therefore, we propose a method to detect potential biases by detecting anomalies in the frequency of operators (e.g., adding two features) suggested by LLMs when engineering new features. Our experiments evaluate the bias of four LLMs, two big frontier and two small open-source models, across 27 tabular datasets. Our results indicate that LLMs are biased toward simple operators, such as addition, and can fail to utilize more complex operators, such as grouping followed by aggregations. Furthermore, the bias can negatively impact the predictive performance when using LLM-generated features. Our results call for mitigating bias when using LLMs for feature engineering.

comment: Preprint

☆ Holon Programming Model -- A Software-Defined Approach for System of Systems

Muhammad Ashfaq, Ahmed R. Sadik, Tommi Mikkonen, Muhammad Waseem, Niko Makitalo

As Systems of Systems evolve into increasingly complex networks, harnessing their collective potential becomes paramount. Traditional SoS engineering approaches lack the necessary programmability to develop third party SoS level behaviors. To address this challenge, we propose a software defined approach to enable flexible and adaptive programming of SoS. We introduce the Holon Programming Model, a software-defined framework designed to meet these needs. The Holon Programming Model empowers developers to design and orchestrate complex system behaviors effectively, as illustrated in our disaster management scenario. This research outlines the Holon Programming Model theoretical underpinnings and practical applications, with the aim of driving further exploration and advancement in the field of software defined SoS

☆ Evaluating Explanations Through LLMs: Beyond Traditional User Studies

Francesco Bombassei De Bona, Gabriele Dominici, Tim Miller, Marc Langheinrich, Martin Gjoreski

As AI becomes fundamental in sectors like healthcare, explainable AI (XAI) tools are essential for trust and transparency. However, traditional user studies used to evaluate these tools are often costly, time consuming, and difficult to scale. In this paper, we explore the use of Large Language Models (LLMs) to replicate human participants to help streamline XAI evaluation. We reproduce a user study comparing counterfactual and causal explanations, replicating human participants with seven LLMs under various settings. Our results show that (i) LLMs can replicate most conclusions from the original study, (ii) different LLMs yield varying levels of alignment in the results, and (iii) experimental factors such as LLM memory and output variability affect alignment with human responses. These initial findings suggest that LLMs could provide a scalable and cost-effective way to simplify qualitative XAI evaluation.

☆ Scaling Robot Policy Learning via Zero-Shot Labeling with Foundation Models

Nils Blank, Moritz Reuss, Marcel Rühle, Ömer Erdinç Yağmurlu, Fabian Wenzel, Oier Mees, Rudolf Lioutikov

A central challenge towards developing robots that can relate human language to their perception and actions is the scarcity of natural language annotations in diverse robot datasets. Moreover, robot policies that follow natural language instructions are typically trained on either templated language or expensive human-labeled instructions, hindering their scalability. To this end, we introduce NILS: Natural language Instruction Labeling for Scalability. NILS automatically labels uncurated, long-horizon robot data at scale in a zero-shot manner without any human intervention. NILS combines pretrained vision-language foundation models in order to detect objects in a scene, detect object-centric changes, segment tasks from large datasets of unlabelled interaction data and ultimately label behavior datasets. Evaluations on BridgeV2, Fractal, and a kitchen play dataset show that NILS can autonomously annotate diverse robot demonstrations of unlabeled and unstructured datasets while alleviating several shortcomings of crowdsourced human annotations, such as low data quality and diversity. We use NILS to label over 115k trajectories obtained from over 430 hours of robot data. We open-source our auto-labeling code and generated annotations on our website: http://robottasklabeling.github.io.

comment: Project Website at https://robottasklabeling.github.io/

☆ Beyond Backpropagation: Optimization with Multi-Tangent Forward Gradients

Katharina Flügel, Daniel Coquelin, Marie Weiel, Achim Streit, Markus Götz

The gradients used to train neural networks are typically computed using backpropagation. While an efficient way to obtain exact gradients, backpropagation is computationally expensive, hinders parallelization, and is biologically implausible. Forward gradients are an approach to approximate the gradients from directional derivatives along random tangents computed by forward-mode automatic differentiation. So far, research has focused on using a single tangent per step. This paper provides an in-depth analysis of multi-tangent forward gradients and introduces an improved approach to combining the forward gradients from multiple tangents based on orthogonal projections. We demonstrate that increasing the number of tangents improves both approximation quality and optimization performance across various tasks.

☆ Escaping the Forest: Sparse Interpretable Neural Networks for Tabular Data

Salvatore Raieli, Abdulrahman Altahhan, Nathalie Jeanray, Stéphane Gerart, Sebastien Vachenc

Tabular datasets are widely used in scientific disciplines such as biology. While these disciplines have already adopted AI methods to enhance their findings and analysis, they mainly use tree-based methods due to their interpretability. At the same time, artificial neural networks have been shown to offer superior flexibility and depth for rich and complex non-tabular problems, but they are falling behind tree-based models for tabular data in terms of performance and interpretability. Although sparsity has been shown to improve the interpretability and performance of ANN models for complex non-tabular datasets, enforcing sparsity structurally and formatively for tabular data before training the model, remains an open question. To address this question, we establish a method that infuses sparsity in neural networks by utilising attention mechanisms to capture the features' importance in tabular datasets. We show that our models, Sparse TABular NET or sTAB-Net with attention mechanisms, are more effective than tree-based models, reaching the state-of-the-art on biological datasets. They further permit the extraction of insights from these datasets and achieve better performance than post-hoc methods like SHAP.

☆ VISAGE: Video Synthesis using Action Graphs for Surgery MICCAI 2024

Yousef Yeganeh, Rachmadio Lazuardi, Amir Shamseddin, Emine Dari, Yash Thirani, Nassir Navab Azade Farshad

Surgical data science (SDS) is a field that analyzes patient data before, during, and after surgery to improve surgical outcomes and skills. However, surgical data is scarce, heterogeneous, and complex, which limits the applicability of existing machine learning methods. In this work, we introduce the novel task of future video generation in laparoscopic surgery. This task can augment and enrich the existing surgical data and enable various applications, such as simulation, analysis, and robot-aided surgery. Ultimately, it involves not only understanding the current state of the operation but also accurately predicting the dynamic and often unpredictable nature of surgical procedures. Our proposed method, VISAGE (VIdeo Synthesis using Action Graphs for Surgery), leverages the power of action scene graphs to capture the sequential nature of laparoscopic procedures and utilizes diffusion models to synthesize temporally coherent video sequences. VISAGE predicts the future frames given only a single initial frame, and the action graph triplets. By incorporating domain-specific knowledge through the action graph, VISAGE ensures the generated videos adhere to the expected visual and motion patterns observed in real laparoscopic procedures. The results of our experiments demonstrate high-fidelity video generation for laparoscopy procedures, which enables various applications in SDS.

comment: Accepted at MICCAI 2024 Embodied AI and Robotics for HealTHcare (EARTH) Workshop

☆ Learning Versatile Skills with Curriculum Masking NeurIPS 2024

Yao Tang, Zhihui Xie, Zichuan Lin, Deheng Ye, Shuai Li

Masked prediction has emerged as a promising pretraining paradigm in offline reinforcement learning (RL) due to its versatile masking schemes, enabling flexible inference across various downstream tasks with a unified model. Despite the versatility of masked prediction, it remains unclear how to balance the learning of skills at different levels of complexity. To address this, we propose CurrMask, a curriculum masking pretraining paradigm for sequential decision making. Motivated by how humans learn by organizing knowledge in a curriculum, CurrMask adjusts its masking scheme during pretraining for learning versatile skills. Through extensive experiments, we show that CurrMask exhibits superior zero-shot performance on skill prompting tasks, goal-conditioned planning tasks, and competitive finetuning performance on offline RL tasks. Additionally, our analysis of training dynamics reveals that CurrMask gradually acquires skills of varying complexity by dynamically adjusting its masking scheme.

comment: NeurIPS 2024 poster, 21 pages, 7 figures

☆ Emotion Recognition with Facial Attention and Objective Activation Functions

Andrzej Miskow, Abdulrahman Altahhan

In this paper, we study the effect of introducing channel and spatial attention mechanisms, namely SEN-Net, ECA-Net, and CBAM, to existing CNN vision-based models such as VGGNet, ResNet, and ResNetV2 to perform the Facial Emotion Recognition task. We show that not only attention can significantly improve the performance of these models but also that combining them with a different activation function can further help increase the performance of these models.

☆ New Insight in Cervical Cancer Diagnosis Using Convolution Neural Network Architecture

Ach. Khozaimi, Wayan Firdaus Mahmudy

The Pap smear is a screening method for early cervical cancer diagnosis. The selection of the right optimizer in the convolutional neural network (CNN) model is key to the success of the CNN in image classification, including the classification of cervical cancer Pap smear images. In this study, stochastic gradient descent (SGD), RMSprop, Adam, AdaGrad, AdaDelta, Adamax, and Nadam optimizers were used to classify cervical cancer Pap smear images from the SipakMed dataset. Resnet-18, Resnet-34, and VGG-16 are the CNN architectures used in this study, and each architecture uses a transfer-learning model. Based on the test results, we conclude that the transfer learning model performs better on all CNNs and optimization techniques and that in the transfer learning model, the optimization has little influence on the training of the model. Adamax, with accuracy values of 72.8% and 66.8%, had the best accuracy for the VGG-16 and Resnet-18 architectures, respectively. Resnet-34 had 54.0%. This is 0.034% lower than Nadam. Overall, Adamax is a suitable optimizer for CNN in cervical cancer classification on Resnet-18, Resnet-34, and VGG-16 architectures. This study provides new insights into the configuration of CNN models for Pap smear image analysis.

☆ FuzzWiz -- Fuzzing Framework for Efficient Hardware Coverage

Deepak Narayan Gadde, Aman Kumar, Djones Lettnin, Sebastian Simon

Ever-increasing design complexity of System-on-Chips (SoCs) led to significant verification challenges. Unlike software, bugs in hardware design are vigorous and eternal i.e., once the hardware is fabricated, it cannot be repaired with any patch. Despite being one of the powerful techniques used in verification, the dynamic random approach cannot give confidence to complex Register Transfer Leve (RTL) designs during the pre-silicon design phase. In particular, achieving coverage targets and exposing bugs is a complicated task with random simulations. In this paper, we leverage an existing testing solution available in the software world known as fuzzing and apply it to hardware verification in order to achieve coverage targets in quick time. We created an automated hardware fuzzing framework FuzzWiz using metamodeling and Python to achieve coverage goals faster. It includes parsing the RTL design module, converting it into C/C++ models, creating generic testbench with assertions, fuzzer-specific compilation, linking, and fuzzing. Furthermore, it is configurable and provides the debug flow if any crash is detected during the fuzzing process. The proposed framework is applied on four IP blocks from Google's OpenTitan chip with various fuzzing engines to show its scalability and compatibility. Our benchmarking results show that we could achieve around 90% of the coverage 10 times faster than traditional simulation regression based approach.

☆ CogSteer: Cognition-Inspired Selective Layer Intervention for Efficient Semantic Steering in Large Language Models

Xintong Wang, Jingheng Pan, Longqin Jiang, Liang Ding, Xingshan Li, Chris Biemann

Despite their impressive capabilities, large language models (LLMs) often lack interpretability and can generate toxic content. While using LLMs as foundation models and applying semantic steering methods are widely practiced, we believe that efficient methods should be based on a thorough understanding of LLM behavior. To this end, we propose using eye movement measures to interpret LLM behavior across layers. We find that LLMs exhibit patterns similar to human gaze across layers and different layers function differently. Inspired by these findings, we introduce a heuristic steering layer selection and apply it to layer intervention methods via fine-tuning and inference. Using language toxification and detoxification as test beds, we demonstrate that our proposed CogSteer methods achieve better results in terms of toxicity scores while efficiently saving 97% of the computational resources and 60% of the training time. Our model-agnostic approach can be adopted into various LLMs, contributing to their interpretability and promoting trustworthiness for safe deployment.

☆ A Data-Driven Odyssey in Solar Vehicles

Do Young Kim, Kyunghyun Kim, Gyeongseop Lee, Niloy Das, Seong-Woo Kim

Solar vehicles, which simultaneously produce and consume energy, require meticulous energy management. However, potential users often feel uncertain about their operation compared to conventional vehicles. This study presents a simulator designed to help users understand long-distance travel in solar vehicles and recognize the importance of proper energy management. By utilizing Google Maps data and weather information, the simulator replicates real-world driving conditions and provides a dashboard displaying vehicle status, updated hourly based on user-inputted speed. Users can explore various speed policy scenarios and receive recommendations for optimal driving strategies. The simulator's effectiveness was validated using the route of the World Solar Challenge (WSC). This research enables users to monitor energy dynamics before a journey, enhancing their understanding of energy management and informing appropriate speed decisions.

☆ Beware of Calibration Data for Pruning Large Language Models

Yixin Ji, Yang Xiang, Juntao Li, Qingrong Xia, Ping Li, Xinyu Duan, Zhefeng Wang, Min Zhang

As large language models (LLMs) are widely applied across various fields, model compression has become increasingly crucial for reducing costs and improving inference efficiency. Post-training pruning is a promising method that does not require resource-intensive iterative training and only needs a small amount of calibration data to assess the importance of parameters. Previous research has primarily focused on designing advanced pruning methods, while different calibration data's impact on pruning performance still lacks systematical exploration. We fill this blank and surprisingly observe that the effects of calibration data even value more than designing advanced pruning strategies, especially for high sparsity. Our preliminary exploration also discloses that using calibration data similar to the training data can yield better performance. As pre-training data is usually inaccessible for advanced LLMs, we further provide a self-generating calibration data synthesis strategy to construct feasible calibration data. We conduct experiments on the recent strong open-source LLMs (e.g., DCLM, and LLaMA-3), and the results show that the proposed method outperforms commonly used calibration data and can effectively enhance strong pruning methods (e.g., Wanda, OWL).

comment: under review

☆ Scalable Random Feature Latent Variable Models

Ying Li, Zhidi Lin, Yuhao Liu, Michael Minyi Zhang, Pablo M. Olmos, Petar M. Djurić

Random feature latent variable models (RFLVMs) represent the state-of-the-art in latent variable models, capable of handling non-Gaussian likelihoods and effectively uncovering patterns in high-dimensional data. However, their heavy reliance on Monte Carlo sampling results in scalability issues which makes it difficult to use these models for datasets with a massive number of observations. To scale up RFLVMs, we turn to the optimization-based variational Bayesian inference (VBI) algorithm which is known for its scalability compared to sampling-based methods. However, implementing VBI for RFLVMs poses challenges, such as the lack of explicit probability distribution functions (PDFs) for the Dirichlet process (DP) in the kernel learning component, and the incompatibility of existing VBI algorithms with RFLVMs. To address these issues, we introduce a stick-breaking construction for DP to obtain an explicit PDF and a novel VBI algorithm called ``block coordinate descent variational inference" (BCD-VI). This enables the development of a scalable version of RFLVMs, or in short, SRFLVM. Our proposed method shows scalability, computational efficiency, superior performance in generating informative latent representations and the ability of imputing missing data across various real-world datasets, outperforming state-of-the-art competitors.

☆ An Adaptive Framework for Generating Systematic Explanatory Answer in Online Q&A Platforms

Ziyang Chen, Xiaobin Wang, Yong Jiang, Jinzhi Liao, Pengjun Xie, Fei Huang, Xiang Zhao

Question Answering (QA) systems face challenges in handling complex questions that require multi-domain knowledge synthesis. The naive RAG models, although effective in information retrieval, struggle with complex questions that require comprehensive and in-depth answers. The pioneering task is defined as explanatory answer generation, which entails handling identified challenges such as the requirement for comprehensive information and logical coherence within the generated context. To address these issues, we refer to systematic thinking theory and propose SynthRAG, an innovative framework designed to enhance QA performance. SynthRAG improves on conventional models by employing adaptive outlines for dynamic content structuring, generating systematic information to ensure detailed coverage, and producing customized answers tailored to specific user inquiries. This structured approach guarantees logical coherence and thorough integration of information, yielding responses that are both insightful and methodically organized. Empirical evaluations underscore SynthRAG's effectiveness, demonstrating its superiority in handling complex questions, overcoming the limitations of naive RAG models, and significantly improving answer quality and depth. Furthermore, an online deployment on the Zhihu platform revealed that SynthRAG's answers achieved notable user engagement, with each response averaging 5.73 upvotes and surpassing the performance of 79.8% of human contributors, highlighting the practical relevance and impact of the proposed framework. Our code is available at https://github.com/czy1999/SynthRAG .

comment: 10 pages, 6 figures

☆ PETAH: Parameter Efficient Task Adaptation for Hybrid Transformers in a resource-limited Context

Maximilian Augustin, Syed Shakib Sarwar, Mostafa Elhoushi, Sai Qian Zhang, Yuecheng Li, Barbara De Salvo

Following their success in natural language processing (NLP), there has been a shift towards transformer models in computer vision. While transformers perform well and offer promising multi-tasking performance, due to their high compute requirements, many resource-constrained applications still rely on convolutional or hybrid models that combine the benefits of convolution and attention layers and achieve the best results in the sub 100M parameter range. Simultaneously, task adaptation techniques that allow for the use of one shared transformer backbone for multiple downstream tasks, resulting in great storage savings at negligible cost in performance, have not yet been adopted for hybrid transformers. In this work, we investigate how to achieve the best task-adaptation performance and introduce PETAH: Parameter Efficient Task Adaptation for Hybrid Transformers. We further combine PETAH adaptation with pruning to achieve highly performant and storage friendly models for multi-tasking. In our extensive evaluation on classification and other vision tasks, we demonstrate that our PETAH-adapted hybrid models outperform established task-adaptation techniques for ViTs while requiring fewer parameters and being more efficient on mobile hardware.

☆ AutoRNet: Automatically Optimizing Heuristics for Robust Network Design via Large Language Models

He Yu, Jing Liu

Achieving robust networks is a challenging problem due to its NP-hard nature and complex solution space. Current methods, from handcrafted feature extraction to deep learning, have made progress but remain rigid, requiring manual design and large labeled datasets. To address these issues, we propose AutoRNet, a framework that integrates large language models (LLMs) with evolutionary algorithms to generate heuristics for robust network design. We design network optimization strategies to provide domain-specific prompts for LLMs, utilizing domain knowledge to generate advanced heuristics. Additionally, we introduce an adaptive fitness function to balance convergence and diversity while maintaining degree distributions. AutoRNet is evaluated on sparse and dense scale-free networks, outperforming current methods by reducing the need for manual design and large datasets.

☆ Mapping the Media Landscape: Predicting Factual Reporting and Political Bias Through Web Interactions

Dairazalia Sánchez-Cortés, Sergio Burdisso, Esaú Villatoro-Tello, Petr Motlicek

Bias assessment of news sources is paramount for professionals, organizations, and researchers who rely on truthful evidence for information gathering and reporting. While certain bias indicators are discernible from content analysis, descriptors like political bias and fake news pose greater challenges. In this paper, we propose an extension to a recently presented news media reliability estimation method that focuses on modeling outlets and their longitudinal web interactions. Concretely, we assess the classification performance of four reinforcement learning strategies on a large news media hyperlink graph. Our experiments, targeting two challenging bias descriptors, factual reporting and political bias, showed a significant performance improvement at the source media level. Additionally, we validate our methods on the CLEF 2023 CheckThat! Lab challenge, outperforming the reported results in both, F1-score and the official MAE metric. Furthermore, we contribute by releasing the largest annotated dataset of news source media, categorized with factual reporting and political bias labels. Our findings suggest that profiling news media sources based on their hyperlink interactions over time is feasible, offering a bird's-eye view of evolving media landscapes.

comment: Accepted to CLEF 2024

☆ MIA-DPO: Multi-Image Augmented Direct Preference Optimization For Large Vision-Language Models

Ziyu Liu, Yuhang Zang, Xiaoyi Dong, Pan Zhang, Yuhang Cao, Haodong Duan, Conghui He, Yuanjun Xiong, Dahua Lin, Jiaqi Wang

Visual preference alignment involves training Large Vision-Language Models (LVLMs) to predict human preferences between visual inputs. This is typically achieved by using labeled datasets of chosen/rejected pairs and employing optimization algorithms like direct preference optimization (DPO). Existing visual alignment methods, primarily designed for single-image scenarios, struggle to effectively handle the complexity of multi-image tasks due to the scarcity of diverse training data and the high cost of annotating chosen/rejected pairs. We present Multi-Image Augmented Direct Preference Optimization (MIA-DPO), a visual preference alignment approach that effectively handles multi-image inputs. MIA-DPO mitigates the scarcity of diverse multi-image training data by extending single-image data with unrelated images arranged in grid collages or pic-in-pic formats, significantly reducing the costs associated with multi-image data annotations. Our observation reveals that attention values of LVLMs vary considerably across different images. We use attention values to identify and filter out rejected responses the model may have mistakenly focused on. Our attention-aware selection for constructing the chosen/rejected pairs without relying on (i) human annotation, (ii) extra data, and (iii) external models or APIs. MIA-DPO is compatible with various architectures and outperforms existing methods on five multi-image benchmarks, achieving an average performance boost of 3.0% on LLaVA-v1.5 and 4.3% on the recent InternLM-XC2.5. Moreover, MIA-DPO has a minimal effect on the model's ability to understand single images.

comment: Project URL: https://github.com/Liuziyu77/MIA-DPO

☆ Markov Chain of Thought for Efficient Mathematical Reasoning

Wen Yang, Kai Fan, Minpeng Liao

Chain of Thought (CoT) of multi-step benefits from the logical structure of the reasoning steps and task-specific actions, significantly enhancing the mathematical reasoning capabilities of large language models. As the prevalence of long CoT, the number of reasoning steps exceeds manageable token limits and leads to higher computational demands. Inspired by the fundamental logic of human cognition, ``derive, then reduce'', we conceptualize the standard multi-step CoT as a novel Markov Chain of Thought (MCoT). In this study, we consider the mathematical reasoning task, defining each reasoning step as text accompanied by a Python code snippet. To facilitate a longer reasoning path, self-correction is enabled through interactions with the code interpreter. Our MCoT aims to compress previous reasoning steps into a simplified question, enabling efficient next-step inference without relying on a lengthy KV cache. In our experiments, we curate the \texttt{MCoTInstruct} dataset, and the empirical results indicate that MCoT not only significantly enhances efficiency but also maintains comparable accuracy. While much remains to be explored, this work paves the way for exploring the long CoT reasoning abilities of LLMs.

comment: Work in progress

☆ LMLPA: Language Model Linguistic Personality Assessment

Jingyao Zheng, Xian Wang, Simo Hosio, Xiaoxian Xu, Lik-Hang Lee

Large Language Models (LLMs) are increasingly used in everyday life and research. One of the most common use cases is conversational interactions, enabled by the language generation capabilities of LLMs. Just as between two humans, a conversation between an LLM-powered entity and a human depends on the personality of the conversants. However, measuring the personality of a given LLM is currently a challenge. This paper introduces the Language Model Linguistic Personality Assessment (LMLPA), a system designed to evaluate the linguistic personalities of LLMs. Our system helps to understand LLMs' language generation capabilities by quantitatively assessing the distinct personality traits reflected in their linguistic outputs. Unlike traditional human-centric psychometrics, the LMLPA adapts a personality assessment questionnaire, specifically the Big Five Inventory, to align with the operational capabilities of LLMs, and also incorporates the findings from previous language-based personality measurement literature. To mitigate sensitivity to the order of options, our questionnaire is designed to be open-ended, resulting in textual answers. Thus, the AI rater is needed to transform ambiguous personality information from text responses into clear numerical indicators of personality traits. Utilising Principal Component Analysis and reliability validations, our findings demonstrate that LLMs possess distinct personality traits that can be effectively quantified by the LMLPA. This research contributes to Human-Computer Interaction and Human-Centered AI, providing a robust framework for future studies to refine AI personality assessments and expand their applications in multiple areas, including education and manufacturing.

☆ Graph Signal Adaptive Message Passing

Yi Yan, Changran Peng, Ercan Engin Kuruoglu

This paper proposes Graph Signal Adaptive Message Passing (GSAMP), a novel message passing method that simultaneously conducts online prediction, missing data imputation, and noise removal on time-varying graph signals. Unlike conventional Graph Signal Processing methods that apply the same filter to the entire graph, the spatiotemporal updates of GSAMP employ a distinct approach that utilizes localized computations at each node. This update is based on an adaptive solution obtained from an optimization problem designed to minimize the discrepancy between observed and estimated values. GSAMP effectively processes real-world, time-varying graph signals under Gaussian and impulsive noise conditions.

☆ Process Supervision-Guided Policy Optimization for Code Generation

Ning Dai, Zheng Wu, Renjie Zheng, Ziyun Wei, Wenlei Shi, Xing Jin, Guanlin Liu, Chen Dun, Liang Huang, Lin Yan

Reinforcement Learning (RL) with unit test feedback has enhanced large language models (LLMs) code generation, but relies on sparse rewards provided only after complete code evaluation, limiting learning efficiency and incremental improvements. When generated code fails all unit tests, no learning signal is received, hindering progress on complex tasks. To address this, we propose a Process Reward Model (PRM) that delivers dense, line-level feedback on code correctness during generation, mimicking human code refinement and providing immediate guidance. We explore various strategies for training PRMs and integrating them into the RL framework, finding that using PRMs both as dense rewards and for value function initialization significantly boosts performance. Our approach increases our in-house LLM's pass rate from 28.2% to 29.8% on LiveCodeBench and from 31.8% to 35.8% on our internal benchmark. Our experimental results highlight the effectiveness of PRMs in enhancing RL-driven code generation, especially for long-horizon scenarios.

comment: 14 pages, 5 figures

☆ From PDFs to Structured Data: Utilizing LLM Analysis in Sports Database Management

Juhani Merilehto

This study investigates the effectiveness of Large Language Models (LLMs) in processing semi-structured data from PDF documents into structured formats, specifically examining their application in updating the Finnish Sports Clubs Database. Through action research methodology, we developed and evaluated an AI-assisted approach utilizing OpenAI's GPT-4 and Anthropic's Claude 3 Opus models to process data from 72 sports federation membership reports. The system achieved a 90% success rate in automated processing, successfully handling 65 of 72 files without errors and converting over 7,900 rows of data. While the initial development time was comparable to traditional manual processing (three months), the implemented system shows potential for reducing future processing time by approximately 90%. Key challenges included handling multilingual content, processing multi-page datasets, and managing extraneous information. The findings suggest that while LLMs demonstrate significant potential for automating semi-structured data processing tasks, optimal results are achieved through a hybrid approach combining AI automation with selective human oversight. This research contributes to the growing body of literature on practical LLM applications in organizational data management and provides insights into the transformation of traditional data processing workflows.

comment: 11 pages, 1 figure

☆ ImDy: Human Inverse Dynamics from Imitated Observations

Xinpeng Liu, Junxuan Liang, Zili Lin, Haowen Hou, Yong-Lu Li, Cewu Lu

Inverse dynamics (ID), which aims at reproducing the driven torques from human kinematic observations, has been a critical tool for gait analysis. However, it is hindered from wider application to general motion due to its limited scalability. Conventional optimization-based ID requires expensive laboratory setups, restricting its availability. To alleviate this problem, we propose to exploit the recently progressive human motion imitation algorithms to learn human inverse dynamics in a data-driven manner. The key insight is that the human ID knowledge is implicitly possessed by motion imitators, though not directly applicable. In light of this, we devise an efficient data collection pipeline with state-of-the-art motion imitation algorithms and physics simulators, resulting in a large-scale human inverse dynamics benchmark as Imitated Dynamics (ImDy). ImDy contains over 150 hours of motion with joint torque and full-body ground reaction force data. With ImDy, we train a data-driven human inverse dynamics solver ImDyS(olver) in a fully supervised manner, which conducts ID and ground reaction force estimation simultaneously. Experiments on ImDy and real-world data demonstrate the impressive competency of ImDyS in human inverse dynamics and ground reaction force estimation. Moreover, the potential of ImDy(-S) as a fundamental motion analysis tool is exhibited with downstream applications. The project page is https://foruck.github.io/ImDy/.

comment: Yong-Lu Li and Cewu Lu are the corresponding authors

☆ Towards Effective Data-Free Knowledge Distillation via Diverse Diffusion Augmentation

Muquan Li, Dongyang Zhang, Tao He, Xiurui Xie, Yuan-Fang Li, Ke Qin

Data-free knowledge distillation (DFKD) has emerged as a pivotal technique in the domain of model compression, substantially reducing the dependency on the original training data. Nonetheless, conventional DFKD methods that employ synthesized training data are prone to the limitations of inadequate diversity and discrepancies in distribution between the synthesized and original datasets. To address these challenges, this paper introduces an innovative approach to DFKD through diverse diffusion augmentation (DDA). Specifically, we revise the paradigm of common data synthesis in DFKD to a composite process through leveraging diffusion models subsequent to data synthesis for self-supervised augmentation, which generates a spectrum of data samples with similar distributions while retaining controlled variations. Furthermore, to mitigate excessive deviation in the embedding space, we introduce an image filtering technique grounded in cosine similarity to maintain fidelity during the knowledge distillation process. Comprehensive experiments conducted on CIFAR-10, CIFAR-100, and Tiny-ImageNet datasets showcase the superior performance of our method across various teacher-student network configurations, outperforming the contemporary state-of-the-art DFKD methods. Code will be available at:https://github.com/SLGSP/DDA.

☆ Integrating Large Language Models for UAV Control in Simulated Environments: A Modular Interaction Approach

Abhishek Phadke, Alihan Hadimlioglu, Tianxing Chu, Chandra N Sekharan

The intersection of LLMs (Large Language Models) and UAV (Unoccupied Aerial Vehicles) technology represents a promising field of research with the potential to enhance UAV capabilities significantly. This study explores the application of LLMs in UAV control, focusing on the opportunities for integrating advanced natural language processing into autonomous aerial systems. By enabling UAVs to interpret and respond to natural language commands, LLMs simplify the UAV control and usage, making them accessible to a broader user base and facilitating more intuitive human-machine interactions. The paper discusses several key areas where LLMs can impact UAV technology, including autonomous decision-making, dynamic mission planning, enhanced situational awareness, and improved safety protocols. Through a comprehensive review of current developments and potential future directions, this study aims to highlight how LLMs can transform UAV operations, making them more adaptable, responsive, and efficient in complex environments. A template development framework for integrating LLMs in UAV control is also described. Proof of Concept results that integrate existing LLM models and popular robotic simulation platforms are demonstrated. The findings suggest that while there are substantial technical and ethical challenges to address, integrating LLMs into UAV control holds promising implications for advancing autonomous aerial systems.

☆ Graphusion: A RAG Framework for Knowledge Graph Construction with a Global Perspective

Rui Yang, Boming Yang, Aosong Feng, Sixun Ouyang, Moritz Blum, Tianwei She, Yuang Jiang, Freddy Lecue, Jinghui Lu, Irene Li

Knowledge Graphs (KGs) are crucial in the field of artificial intelligence and are widely used in downstream tasks, such as question-answering (QA). The construction of KGs typically requires significant effort from domain experts. Large Language Models (LLMs) have recently been used for Knowledge Graph Construction (KGC). However, most existing approaches focus on a local perspective, extracting knowledge triplets from individual sentences or documents, missing a fusion process to combine the knowledge in a global KG. This work introduces Graphusion, a zero-shot KGC framework from free text. It contains three steps: in Step 1, we extract a list of seed entities using topic modeling to guide the final KG includes the most relevant entities; in Step 2, we conduct candidate triplet extraction using LLMs; in Step 3, we design the novel fusion module that provides a global view of the extracted knowledge, incorporating entity merging, conflict resolution, and novel triplet discovery. Results show that Graphusion achieves scores of 2.92 and 2.37 out of 3 for entity extraction and relation recognition, respectively. Moreover, we showcase how Graphusion could be applied to the Natural Language Processing (NLP) domain and validate it in an educational scenario. Specifically, we introduce TutorQA, a new expert-verified benchmark for QA, comprising six tasks and a total of 1,200 QA pairs. Using the Graphusion-constructed KG, we achieve a significant improvement on the benchmark, for example, a 9.2% accuracy improvement on sub-graph completion.

comment: arXiv admin note: substantial text overlap with arXiv:2407.10794

☆ Challenge on Sound Scene Synthesis: Evaluating Text-to-Audio Generation NeurIPS 2024

Junwon Lee, Modan Tailleur, Laurie M. Heller, Keunwoo Choi, Mathieu Lagrange, Brian McFee, Keisuke Imoto, Yuki Okamoto

Despite significant advancements in neural text-to-audio generation, challenges persist in controllability and evaluation. This paper addresses these issues through the Sound Scene Synthesis challenge held as part of the Detection and Classification of Acoustic Scenes and Events 2024. We present an evaluation protocol combining objective metric, namely Fr\'echet Audio Distance, with perceptual assessments, utilizing a structured prompt format to enable diverse captions and effective evaluation. Our analysis reveals varying performance across sound categories and model architectures, with larger models generally excelling but innovative lightweight approaches also showing promise. The strong correlation between objective metrics and human ratings validates our evaluation approach. We discuss outcomes in terms of audio quality, controllability, and architectural considerations for text-to-audio synthesizers, providing direction for future research.

comment: accepted to NeurIPS 2024 Workshop: Audio Imagination

☆ Exploring Tokenization Methods for Multitrack Sheet Music Generation

Yashan Wang, Shangda Wu, Xingjian Du, Maosong Sun

This study explores the tokenization of multitrack sheet music in ABC notation, introducing two methods--bar-stream and line-stream patching. We compare these methods against existing techniques, including bar patching, byte patching, and Byte Pair Encoding (BPE). In terms of both computational efficiency and the musicality of the generated compositions, experimental results show that bar-stream patching performs best overall compared to the others, which makes it a promising tokenization strategy for sheet music generation.

comment: 3 pages, 1 figure, 1 table

☆ Bonsai: Gradient-free Graph Distillation for Node Classification

Mridul Gupta, Samyak Jain, Vansh Ramani, Hariprasad Kodamana, Sayan Ranu

Graph distillation has emerged as a promising avenue to enable scalable training of GNNs by compressing the training dataset while preserving essential graph characteristics. Our study uncovers significant shortcomings in current graph distillation techniques. First, the majority of the algorithms paradoxically require training on the full dataset to perform distillation. Second, due to their gradient-emulating approach, these methods require fresh distillation for any change in hyperparameters or GNN architecture, limiting their flexibility and reusability. Finally, they fail to achieve substantial size reduction due to synthesizing fully-connected, edge-weighted graphs. To address these challenges, we present Bonsai, a novel graph distillation method empowered by the observation that \textit{computation trees} form the fundamental processing units of message-passing GNNs. Bonsai distills datasets by encoding a careful selection of \textit{exemplar} trees that maximize the representation of all computation trees in the training set. This unique approach imparts Bonsai as the first linear-time, model-agnostic graph distillation algorithm for node classification that outperforms existing baselines across $6$ real-world datasets on accuracy, while being $22$ times faster on average. Bonsai is grounded in rigorous mathematical guarantees on the adopted approximation strategies making it robust to GNN architectures, datasets, and parameters.

☆ Real-time Vehicle-to-Vehicle Communication Based Network Cooperative Control System through Distributed Database and Multimodal Perception: Demonstrated in Crossroads

Xinwen Zhu, Zihao Li, Yuxuan Jiang, Jiazhen Xu, Jie Wang, Xuyang Bai

The autonomous driving industry is rapidly advancing, with Vehicle-to-Vehicle (V2V) communication systems highlighting as a key component of enhanced road safety and traffic efficiency. This paper introduces a novel Real-time Vehicle-to-Vehicle Communication Based Network Cooperative Control System (VVCCS), designed to revolutionize macro-scope traffic planning and collision avoidance in autonomous driving. Implemented on Quanser Car (Qcar) hardware platform, our system integrates the distributed databases into individual autonomous vehicles and an optional central server. We also developed a comprehensive multi-modal perception system with multi-objective tracking and radar sensing. Through a demonstration within a physical crossroad environment, our system showcases its potential to be applied in congested and complex urban environments.

comment: ICICT 2024, 18 pages

☆ Differentially Private Learning Needs Better Model Initialization and Self-Distillation

Ivoline C. Ngong, Joseph P. Near, Niloofar Mireshghallah

Differentially private SGD (DPSGD) enables privacy-preserving training of language models, but often reduces utility, diversity, and linguistic quality. We introduce DPRefine, a three-phase method that initializes a model using data synthesis from a small pre-trained LM with rigorous filtering, applies DP finetuning on private data, and performs self-distillation to refine outputs. This approach significantly outperforms vanilla DPSGD, with AlpacaEval preferring DPRefine's generations in 78.4% of cases across all datasets. Our analysis reveals that DPRefine reduces linguistic errors in generated text by 84.0%, mitigating grammar and spelling errors, commonly associated with DPSGD. It also reduces inconsistencies of non-private models, such as hallucinated details and misattributed quotes. We find that small models like GPT-2 can be effective for initialization and distillation, highlighting their potential in enabling scalable and efficient deployment of privacy-preserving language.

comment: 18 pages

☆ CLR-Bench: Evaluating Large Language Models in College-level Reasoning

Junnan Dong, Zijin Hong, Yuanchen Bei, Feiran Huang, Xinrun Wang, Xiao Huang

Large language models (LLMs) have demonstrated their remarkable performance across various language understanding tasks. While emerging benchmarks have been proposed to evaluate LLMs in various domains such as mathematics and computer science, they merely measure the accuracy in terms of the final prediction on multi-choice questions. However, it remains insufficient to verify the essential understanding of LLMs given a chosen choice. To fill this gap, we present CLR-Bench to comprehensively evaluate the LLMs in complex college-level reasoning. Specifically, (i) we prioritize 16 challenging college disciplines in computer science and artificial intelligence. The dataset contains 5 types of questions, while each question is associated with detailed explanations from experts. (ii) To quantify a fair evaluation of LLMs' reasoning ability, we formalize the criteria with two novel metrics. Q$\rightarrow$A is utilized to measure the performance of direct answer prediction, and Q$\rightarrow$AR effectively considers the joint ability to answer the question and provide rationale simultaneously. Extensive experiments are conducted with 40 LLMs over 1,018 discipline-specific questions. The results demonstrate the key insights that LLMs, even the best closed-source LLM, i.e., GPT-4 turbo, tend to `guess' the college-level answers. It shows a dramatic decrease in accuracy from 63.31% Q$\rightarrow$A to 39.00% Q$\rightarrow$AR, indicating an unsatisfactory reasoning ability.

comment: 18 pages, 6 figures, dataset and evaluation framework will be opensourced

☆ FairDgcl: Fairness-aware Recommendation with Dynamic Graph Contrastive Learning

Wei Chen, Meng Yuan, Zhao Zhang, Ruobing Xie, Fuzhen Zhuang, Deqing Wang, Rui Liu

As trustworthy AI continues to advance, the fairness issue in recommendations has received increasing attention. A recommender system is considered unfair when it produces unequal outcomes for different user groups based on user-sensitive attributes (e.g., age, gender). Some researchers have proposed data augmentation-based methods aiming at alleviating user-level unfairness by altering the skewed distribution of training data among various user groups. Despite yielding promising results, they often rely on fairness-related assumptions that may not align with reality, potentially reducing the data quality and negatively affecting model effectiveness. To tackle this issue, in this paper, we study how to implement high-quality data augmentation to improve recommendation fairness. Specifically, we propose FairDgcl, a dynamic graph adversarial contrastive learning framework aiming at improving fairness in recommender system. First, FairDgcl develops an adversarial contrastive network with a view generator and a view discriminator to learn generating fair augmentation strategies in an adversarial style. Then, we propose two dynamic, learnable models to generate contrastive views within contrastive learning framework, which automatically fine-tune the augmentation strategies. Meanwhile, we theoretically show that FairDgcl can simultaneously generate enhanced representations that possess both fairness and accuracy. Lastly, comprehensive experiments conducted on four real-world datasets demonstrate the effectiveness of the proposed FairDgcl.

comment: 12 pages, submitted to TKDE

☆ ProtoLens: Advancing Prototype Learning for Fine-Grained Interpretability in Text Classification

Bowen Wei, Ziwei Zhu

Deep neural networks have achieved remarkable performance in various text-based tasks but often lack interpretability, making them less suitable for applications where transparency is critical. To address this, we propose ProtoLens, a novel prototype-based model that provides fine-grained, sub-sentence level interpretability for text classification. ProtoLens uses a Prototype-aware Span Extraction module to identify relevant text spans associated with learned prototypes and a Prototype Alignment mechanism to ensure prototypes are semantically meaningful throughout training. By aligning the prototype embeddings with human-understandable examples, ProtoLens provides interpretable predictions while maintaining competitive accuracy. Extensive experiments demonstrate that ProtoLens outperforms both prototype-based and non-interpretable baselines on multiple text classification benchmarks. Code and data are available at \url{https://anonymous.4open.science/r/ProtoLens-CE0B/}.

☆ Primal-Dual Spectral Representation for Off-policy Evaluation

Yang Hu, Tianyi Chen, Na Li, Kai Wang, Bo Dai

Off-policy evaluation (OPE) is one of the most fundamental problems in reinforcement learning (RL) to estimate the expected long-term payoff of a given target policy with only experiences from another behavior policy that is potentially unknown. The distribution correction estimation (DICE) family of estimators have advanced the state of the art in OPE by breaking the curse of horizon. However, the major bottleneck of applying DICE estimators lies in the difficulty of solving the saddle-point optimization involved, especially with neural network implementations. In this paper, we tackle this challenge by establishing a linear representation of value function and stationary distribution correction ratio, i.e., primal and dual variables in the DICE framework, using the spectral decomposition of the transition operator. Such primal-dual representation not only bypasses the non-convex non-concave optimization in vanilla DICE, therefore enabling an computational efficient algorithm, but also paves the way for more efficient utilization of historical data. We highlight that our algorithm, SpectralDICE, is the first to leverage the linear representation of primal-dual variables that is both computation and sample efficient, the performance of which is supported by a rigorous theoretical sample complexity guarantee and a thorough empirical evaluation on various benchmarks.

comment: 29 pages, 5 figures

☆ Responsible Multilingual Large Language Models: A Survey of Development, Applications, and Societal Impact

Junhua Liu, Bin Fu

Multilingual Large Language Models (MLLMs) represent a pivotal advancement in democratizing artificial intelligence across linguistic boundaries. While theoretical foundations are well-established, practical implementation guidelines remain scattered. This work bridges this gap by providing a comprehensive end-to-end framework for developing and deploying MLLMs in production environments. We make three distinctive contributions: First, we present an actionable pipeline from data pre-processing through deployment, integrating insights from academic research and industrial applications. Second, using Llama2 as a case study, we provide detailed optimization strategies for enhancing multilingual capabilities, including curriculum learning approaches for balancing high-resource and low-resource languages, tokenization strategies, and effective sampling methods. Third, we offer an interdisciplinary analysis that considers technical, linguistic, and cultural perspectives in MLLM development. Our findings reveal critical challenges in supporting linguistic diversity, with 88.38% of world languages categorized as low-resource, affecting over a billion speakers. We examine practical solutions through real-world applications in customer service, search engines, and machine translation. By synthesizing theoretical frameworks with production-ready implementation strategies, this survey provides essential guidance for practitioners and researchers working to develop more inclusive and effective multilingual AI systems.

☆ Bridging Swarm Intelligence and Reinforcement Learning

Karthik Soma, Yann Bouteiller, Heiko Hamann, Giovanni Beltrame

Swarm intelligence (SI) explores how large groups of simple individuals (e.g., insects, fish, birds) collaborate to produce complex behaviors, exemplifying that the whole is greater than the sum of its parts. A fundamental task in SI is Collective Decision-Making (CDM), where a group selects the best option among several alternatives, such as choosing an optimal foraging site. In this work, we demonstrate a theoretical and empirical equivalence between CDM and single-agent reinforcement learning (RL) in multi-armed bandit problems, utilizing concepts from opinion dynamics, evolutionary game theory, and RL. This equivalence bridges the gap between SI and RL and leads us to introduce a novel abstract RL update rule called Maynard-Cross Learning. Additionally, it provides a new population-based perspective on common RL practices like learning rate adjustment and batching. Our findings enable cross-disciplinary fertilization between RL and SI, allowing techniques from one field to enhance the understanding and methodologies of the other.

☆ Time and Frequency Synergy for Source-Free Time-Series Domain Adaptations

Muhammad Tanzil Furqon, Mahardhika Pratama, Ary Mazharuddin Shiddiqi, Lin Liu, Habibullah Habibullah, Kutluyil Dogancay

The issue of source-free time-series domain adaptations still gains scarce research attentions. On the other hand, existing approaches rely solely on time-domain features ignoring frequency components providing complementary information. This paper proposes Time Frequency Domain Adaptation (TFDA), a method to cope with the source-free time-series domain adaptation problems. TFDA is developed with a dual branch network structure fully utilizing both time and frequency features in delivering final predictions. It induces pseudo-labels based on a neighborhood concept where predictions of a sample group are aggregated to generate reliable pseudo labels. The concept of contrastive learning is carried out in both time and frequency domains with pseudo label information and a negative pair exclusion strategy to make valid neighborhood assumptions. In addition, the time-frequency consistency technique is proposed using the self-distillation strategy while the uncertainty reduction strategy is implemented to alleviate uncertainties due to the domain shift problem. Last but not least, the curriculum learning strategy is integrated to combat noisy pseudo labels. Our experiments demonstrate the advantage of our approach over prior arts with noticeable margins in benchmark problems.

☆ Congestion Forecast for Trains with Railroad-Graph-based Semi-Supervised Learning using Sparse Passenger Reports SP

Soto Anno, Kota Tsubouchi, Masamichi Shimosaka

Forecasting rail congestion is crucial for efficient mobility in transport systems. We present rail congestion forecasting using reports from passengers collected through a transit application. Although reports from passengers have received attention from researchers, ensuring a sufficient volume of reports is challenging due to passenger's reluctance. The limited number of reports results in the sparsity of the congestion label, which can be an issue in building a stable prediction model. To address this issue, we propose a semi-supervised method for congestion forecasting for trains, or SURCONFORT. Our key idea is twofold: firstly, we adopt semi-supervised learning to leverage sparsely labeled data and many unlabeled data. Secondly, in order to complement the unlabeled data from nearby stations, we design a railway network-oriented graph and apply the graph to semi-supervised graph regularization. Empirical experiments with actual reporting data show that SURCONFORT improved the forecasting performance by 14.9% over state-of-the-art methods under the label sparsity.

comment: Accepted in ACM SIGSPATIAL 2024

☆ Mitigating Graph Covariate Shift via Score-based Out-of-distribution Augmentation

Bohan Wang, Yurui Chang, Lu Lin

Distribution shifts between training and testing datasets significantly impair the model performance on graph learning. A commonly-taken causal view in graph invariant learning suggests that stable predictive features of graphs are causally associated with labels, whereas varying environmental features lead to distribution shifts. In particular, covariate shifts caused by unseen environments in test graphs underscore the critical need for out-of-distribution (OOD) generalization. Existing graph augmentation methods designed to address the covariate shift often disentangle the stable and environmental features in the input space, and selectively perturb or mixup the environmental features. However, such perturbation-based methods heavily rely on an accurate separation of stable and environmental features, and their exploration ability is confined to existing environmental features in the training distribution. To overcome these limitations, we introduce a novel approach using score-based graph generation strategies that synthesize unseen environmental features while preserving the validity and stable features of overall graph patterns. Our comprehensive empirical evaluations demonstrate the enhanced effectiveness of our method in improving graph OOD generalization.

comment: 17 pages, 5 figures, 4 tables

☆ An Ontology-Enabled Approach For User-Centered and Knowledge-Enabled Explanations of AI Systems

Shruthi Chari

Explainable Artificial Intelligence (AI) focuses on helping humans understand the working of AI systems or their decisions and has been a cornerstone of AI for decades. Recent research in explainability has focused on explaining the workings of AI models or model explainability. There have also been several position statements and review papers detailing the needs of end-users for user-centered explainability but fewer implementations. Hence, this thesis seeks to bridge some gaps between model and user-centered explainability. We create an explanation ontology (EO) to represent literature-derived explanation types via their supporting components. We implement a knowledge-augmented question-answering (QA) pipeline to support contextual explanations in a clinical setting. Finally, we are implementing a system to combine explanations from different AI methods and data modalities. Within the EO, we can represent fifteen different explanation types, and we have tested these representations in six exemplar use cases. We find that knowledge augmentations improve the performance of base large language models in the contextualized QA, and the performance is variable across disease groups. In the same setting, clinicians also indicated that they prefer to see actionability as one of the main foci in explanations. In our explanations combination method, we plan to use similarity metrics to determine the similarity of explanations in a chronic disease detection setting. Overall, through this thesis, we design methods that can support knowledge-enabled explanations across different use cases, accounting for the methods in today's AI era that can generate the supporting components of these explanations and domain knowledge sources that can enhance them.

comment: Doctoral dissertation. Some chapters appeared as individual papers - arXiv:2302.05752 is one such chapters

☆ Learning Fair and Preferable Allocations through Neural Network

Ryota Maruo, Koh Takeuchi, Hisashi Kashima

The fair allocation of indivisible resources is a fundamental problem. Existing research has developed various allocation mechanisms or algorithms to satisfy different fairness notions. For example, round robin (RR) was proposed to meet the fairness criterion known as envy-freeness up to one good (EF1). Expert algorithms without mathematical formulations are used in real-world resource allocation problems to find preferable outcomes for users. Therefore, we aim to design mechanisms that strictly satisfy good properties with replicating expert knowledge. However, this problem is challenging because such heuristic rules are often difficult to formalize mathematically, complicating their integration into theoretical frameworks. Additionally, formal algorithms struggle to find preferable outcomes, and directly replicating these implicit rules can result in unfair allocations because human decision-making can introduce biases. In this paper, we aim to learn implicit allocation mechanisms from examples while strictly satisfying fairness constraints, specifically focusing on learning EF1 allocation mechanisms through supervised learning on examples of reported valuations and corresponding allocation outcomes produced by implicit rules. To address this, we developed a neural RR (NRR), a novel neural network that parameterizes RR. NRR is built from a differentiable relaxation of RR and can be trained to learn the agent ordering used for RR. We conducted experiments to learn EF1 allocation mechanisms from examples, demonstrating that our method outperforms baselines in terms of the proximity of predicted allocations and other metrics.

☆ Mechanisms of Symbol Processing for In-Context Learning in Transformer Networks

Paul Smolensky, Roland Fernandez, Zhenghao Herbert Zhou, Mattia Opper, Jianfeng Gao

Large Language Models (LLMs) have demonstrated impressive abilities in symbol processing through in-context learning (ICL). This success flies in the face of decades of predictions that artificial neural networks cannot master abstract symbol manipulation. We seek to understand the mechanisms that can enable robust symbol processing in transformer networks, illuminating both the unanticipated success, and the significant limitations, of transformers in symbol processing. Borrowing insights from symbolic AI on the power of Production System architectures, we develop a high-level language, PSL, that allows us to write symbolic programs to do complex, abstract symbol processing, and create compilers that precisely implement PSL programs in transformer networks which are, by construction, 100% mechanistically interpretable. We demonstrate that PSL is Turing Universal, so the work can inform the understanding of transformer ICL in general. The type of transformer architecture that we compile from PSL programs suggests a number of paths for enhancing transformers' capabilities at symbol processing. (Note: The first section of the paper gives an extended synopsis of the entire paper.)

comment: 101 pages (including 30 pages of Appendices), 18 figures

☆ Unsupervised Domain Adaptation for Action Recognition via Self-Ensembling and Conditional Embedding Alignment

Indrajeet Ghosh, Garvit Chugh, Abu Zaher Md Faridee, Nirmalya Roy

Recent advancements in deep learning-based wearable human action recognition (wHAR) have improved the capture and classification of complex motions, but adoption remains limited due to the lack of expert annotations and domain discrepancies from user variations. Limited annotations hinder the model's ability to generalize to out-of-distribution samples. While data augmentation can improve generalizability, unsupervised augmentation techniques must be applied carefully to avoid introducing noise. Unsupervised domain adaptation (UDA) addresses domain discrepancies by aligning conditional distributions with labeled target samples, but vanilla pseudo-labeling can lead to error propagation. To address these challenges, we propose $\mu$DAR, a novel joint optimization architecture comprised of three functions: (i) consistency regularizer between augmented samples to improve model classification generalizability, (ii) temporal ensemble for robust pseudo-label generation and (iii) conditional distribution alignment to improve domain generalizability. The temporal ensemble works by aggregating predictions from past epochs to smooth out noisy pseudo-label predictions, which are then used in the conditional distribution alignment module to minimize kernel-based class-wise conditional maximum mean discrepancy ($k$CMMD) between the source and target feature space to learn a domain invariant embedding. The consistency-regularized augmentations ensure that multiple augmentations of the same sample share the same labels; this results in (a) strong generalization with limited source domain samples and (b) consistent pseudo-label generation in target samples. The novel integration of these three modules in $\mu$DAR results in a range of $\approx$ 4-12% average macro-F1 score improvement over six state-of-the-art UDA methods in four benchmark wHAR datasets

comment: This work has been accepted to the Proceedings of the IEEE International Conference on Data Mining, 2024

☆ AI, Global Governance, and Digital Sovereignty

Swati Srivastava, Justin Bullock

This essay examines how Artificial Intelligence (AI) systems are becoming more integral to international affairs by affecting how global governors exert power and pursue digital sovereignty. We first introduce a taxonomy of multifaceted AI payoffs for governments and corporations related to instrumental, structural, and discursive power in the domains of violence, markets, and rights. We next leverage different institutional and practice perspectives on sovereignty to assess how digital sovereignty is variously implicated in AI-empowered global governance. States both seek sovereign control over AI infrastructures in the institutional approach, while establishing sovereign competence through AI infrastructures in the practice approach. Overall, we present the digital sovereignty stakes of AI as related to entanglements of public and private power. Rather than foreseeing technology companies as replacing states, we argue that AI systems will embed in global governance to create dueling dynamics of public/private cooperation and contestation. We conclude with sketching future directions for IR research on AI and global governance.

comment: 21 pages, 2 tables

♻ ☆ Non-myopic Generation of Language Model for Reasoning and Planning

Chang Ma, Haiteng Zhao, Junlei Zhang, Junxian He, Lingpeng Kong

Large Language Models have demonstrated remarkable abilities in reasoning and planning by breaking down complex problems into sequential steps. Despite their success in various domains like mathematical problem-solving and coding, LLMs face challenges in ensuring reliable and optimal planning due to their inherent myopic nature of autoregressive decoding. This paper revisits LLM reasoning from an optimal-control perspective, proposing a novel method, Predictive-Decoding, that leverages Model Predictive Control to enhance planning accuracy. By re-weighting LLM distributions based on foresight trajectories, Predictive-Decoding aims to mitigate early errors and promote non-myopic planning. Our experiments show significant improvements in a wide range of tasks for math, coding, and agents. Furthermore, Predictive-Decoding demonstrates computational efficiency, outperforming search baselines with reduced computational resources. This study provides insights into optimizing LLM planning capabilities.

♻ ☆ GeoCode-GPT: A Large Language Model for Geospatial Code Generation Tasks

Shuyang Hou, Zhangxiao Shen, Anqi Zhao, Jianyuan Liang, Zhipeng Gui, Xuefeng Guan, Rui Li, Huayi Wu

The increasing demand for spatiotemporal data and modeling tasks in geosciences has made geospatial code generation technology a critical factor in enhancing productivity. Although large language models (LLMs) have demonstrated potential in code generation tasks, they often encounter issues such as refusal to code or hallucination in geospatial code generation due to a lack of domain-specific knowledge and code corpora. To address these challenges, this paper presents and open-sources the GeoCode-PT and GeoCode-SFT corpora, along with the GeoCode-Eval evaluation dataset. Additionally, by leveraging QLoRA and LoRA for pretraining and fine-tuning, we introduce GeoCode-GPT-7B, the first LLM focused on geospatial code generation, fine-tuned from Code Llama-7B. Furthermore, we establish a comprehensive geospatial code evaluation framework, incorporating option matching, expert validation, and prompt engineering scoring for LLMs, and systematically evaluate GeoCode-GPT-7B using the GeoCode-Eval dataset. Experimental results show that GeoCode-GPT outperforms other models in multiple-choice accuracy by 9.1% to 32.1%, in code summarization ability by 1.7% to 25.4%, and in code generation capability by 1.2% to 25.1%. This paper provides a solution and empirical validation for enhancing LLMs' performance in geospatial code generation, extends the boundaries of domain-specific model applications, and offers valuable insights into unlocking their potential in geospatial code generation.

♻ ☆ AskBeacon -- Performing genomic data exchange and analytics with natural language

Anuradha Wickramarachchi, Shakila Tonni, Sonali Majumdar, Sarvnaz Karimi, Sulev Kõks, Brendan Hosking, Jordi Rambla, Natalie A. Twine, Yatish Jain, Denis C. Bauer

Enabling clinicians and researchers to directly interact with global genomic data resources by removing technological barriers is vital for medical genomics. AskBeacon enables Large Language Models to be applied to securely shared cohorts via the GA4GH Beacon protocol. By simply "asking" Beacon, actionable insights can be gained, analyzed and made publication-ready.

♻ ☆ LLMScan: Causal Scan for LLM Misbehavior Detection

Mengdi Zhang, Kai Kiat Goh, Peixin Zhang, Jun Sun

Despite the success of Large Language Models (LLMs) across various fields, their potential to generate untruthful, biased and harmful responses poses significant risks, particularly in critical applications. This highlights the urgent need for systematic methods to detect and prevent such misbehavior. While existing approaches target specific issues such as harmful responses, this work introduces LLMScan, an innovative LLM monitoring technique based on causality analysis, offering a comprehensive solution. LLMScan systematically monitors the inner workings of an LLM through the lens of causal inference, operating on the premise that the LLM's `brain' behaves differently when misbehaving. By analyzing the causal contributions of the LLM's input tokens and transformer layers, LLMScan effectively detects misbehavior. Extensive experiments across various tasks and models reveal clear distinctions in the causal distributions between normal behavior and misbehavior, enabling the development of accurate, lightweight detectors for a variety of misbehavior detection tasks.

♻ ☆ Diffusion-Reward Adversarial Imitation Learning

Chun-Mao Lai, Hsiang-Chun Wang, Ping-Chun Hsieh, Yu-Chiang Frank Wang, Min-Hung Chen, Shao-Hua Sun

Imitation learning aims to learn a policy from observing expert demonstrations without access to reward signals from environments. Generative adversarial imitation learning (GAIL) formulates imitation learning as adversarial learning, employing a generator policy learning to imitate expert behaviors and discriminator learning to distinguish the expert demonstrations from agent trajectories. Despite its encouraging results, GAIL training is often brittle and unstable. Inspired by the recent dominance of diffusion models in generative modeling, we propose Diffusion-Reward Adversarial Imitation Learning (DRAIL), which integrates a diffusion model into GAIL, aiming to yield more robust and smoother rewards for policy learning. Specifically, we propose a diffusion discriminative classifier to construct an enhanced discriminator, and design diffusion rewards based on the classifier's output for policy learning. Extensive experiments are conducted in navigation, manipulation, and locomotion, verifying DRAIL's effectiveness compared to prior imitation learning methods. Moreover, additional experimental results demonstrate the generalizability and data efficiency of DRAIL. Visualized learned reward functions of GAIL and DRAIL suggest that DRAIL can produce more robust and smoother rewards. Project page: https://nturobotlearninglab.github.io/DRAIL/

♻ ☆ Pruning By Explaining Revisited: Optimizing Attribution Methods to Prune CNNs and Transformers ECCV 2024

Sayed Mohammad Vakilzadeh Hatefi, Maximilian Dreyer, Reduan Achtibat, Thomas Wiegand, Wojciech Samek, Sebastian Lapuschkin

To solve ever more complex problems, Deep Neural Networks are scaled to billions of parameters, leading to huge computational costs. An effective approach to reduce computational requirements and increase efficiency is to prune unnecessary components of these often over-parameterized networks. Previous work has shown that attribution methods from the field of eXplainable AI serve as effective means to extract and prune the least relevant network components in a few-shot fashion. We extend the current state by proposing to explicitly optimize hyperparameters of attribution methods for the task of pruning, and further include transformer-based networks in our analysis. Our approach yields higher model compression rates of large transformer- and convolutional architectures (VGG, ResNet, ViT) compared to previous works, while still attaining high performance on ImageNet classification tasks. Here, our experiments indicate that transformers have a higher degree of over-parameterization compared to convolutional neural networks. Code is available at https://github.com/erfanhatefi/Pruning-by-eXplaining-in-PyTorch.

comment: Accepted as a workshop paper at ECCV 2024, 26 pages (11 pages manuscript, 3 pages references, 12 pages appendix)

♻ ☆ Correlated Proxies: A New Definition and Improved Mitigation for Reward Hacking

Cassidy Laidlaw, Shivam Singhal, Anca Dragan

Because it is difficult to precisely specify complex objectives, reinforcement learning policies are often optimized using flawed proxy rewards that seem to capture the true objective. However, optimizing proxy rewards frequently leads to reward hacking: the optimized reward function ceases to be a good proxy, and the resulting policy performs poorly with respect to the unspecified true reward. Principled solutions to reward hacking have been impeded by the lack of a good definition for the problem. To address this, we introduce a definition of reward hacking based on the correlation between proxy and true rewards for states and actions seen by a "base policy" that breaks down under optimization. We show that this definition captures reward hacking behavior across several realistic settings, including in reinforcement learning from human feedback (RLHF). We then show theoretically that regularization to the base policy can effectively prevent reward hacking. While current RLHF approaches apply a KL penalty between the action distributions of policies, our theory suggests that it is more effective to regularize using the $\chi^2$ divergence between the policies' occupancy measures. We intuitively show why this type of regularization is superior and demonstrate that it better mitigates reward hacking in practice across four realistic domains, including RLHF for LLMs. Our code is available at https://github.com/cassidylaidlaw/orpo.

♻ ☆ Physical Reasoning and Object Planning for Household Embodied Agents

Ayush Agrawal, Raghav Prabhakar, Anirudh Goyal, Dianbo Liu

In this study, we explore the sophisticated domain of task planning for robust household embodied agents, with a particular emphasis on the intricate task of selecting substitute objects. We introduce the CommonSense Object Affordance Task (COAT), a novel framework designed to analyze reasoning capabilities in commonsense scenarios. This approach is centered on understanding how these agents can effectively identify and utilize alternative objects when executing household tasks, thereby offering insights into the complexities of practical decision-making in real-world environments. Drawing inspiration from factors affecting human decision-making, we explore how large language models tackle this challenge through four meticulously crafted commonsense question-and-answer datasets featuring refined rules and human annotations. Our evaluation of state-of-the-art language models on these datasets sheds light on three pivotal considerations: 1) aligning an object's inherent utility with the task at hand, 2) navigating contextual dependencies (societal norms, safety, appropriateness, and efficiency), and 3) accounting for the current physical state of the object. To maintain accessibility, we introduce five abstract variables reflecting an object's physical condition, modulated by human insights, to simulate diverse household scenarios. Our contributions include insightful human preference mappings for all three factors and four extensive QA datasets (2K, 15k, 60k, 70K questions) probing the intricacies of utility dependencies, contextual dependencies and object physical states. The datasets, along with our findings, are accessible at: https://github.com/Ayush8120/COAT. This research not only advances our understanding of physical commonsense reasoning in language models but also paves the way for future improvements in household agent intelligence.

comment: Journal: TMLR(May/2024) Total: 39 pages (17 pages main content, 15 Figures)

♻ ☆ MADial-Bench: Towards Real-world Evaluation of Memory-Augmented Dialogue Generation NAACL 2025

Junqing He, Liang Zhu, Rui Wang, Xi Wang, Reza Haffari, Jiaxing Zhang

Long-term memory is important for chatbots and dialogue systems (DS) to create consistent and human-like conversations, evidenced by numerous developed memory-augmented DS (MADS). To evaluate the effectiveness of such MADS, existing commonly used evaluation metrics, like retrieval accuracy and perplexity (PPL), mainly focus on query-oriented factualness and language quality assessment. However, these metrics often lack practical value. Moreover, the evaluation dimensions are insufficient for human-like assessment in DS. Regarding memory-recalling paradigms, current evaluation schemes only consider passive memory retrieval while ignoring diverse memory recall with rich triggering factors, e.g., emotions and surroundings, which can be essential in emotional support scenarios. To bridge the gap, we construct a novel Memory-Augmented Dialogue Benchmark (MADail-Bench) covering various memory-recalling paradigms based on cognitive science and psychology theories. The benchmark assesses two tasks separately: memory retrieval and memory recognition with the incorporation of both passive and proactive memory recall data. We introduce new scoring criteria to the evaluation, including memory injection, emotion support (ES) proficiency, and intimacy, to comprehensively assess generated responses. Results from cutting-edge embedding models and large language models on this benchmark indicate the potential for further advancement. Extensive testing further reveals correlations between memory injection, ES proficiency, and intimacy.

comment: Submitted to NAACL 2025

♻ ☆ Conditional Language Policy: A General Framework for Steerable Multi-Objective Finetuning EMNLP 2024

Kaiwen Wang, Rahul Kidambi, Ryan Sullivan, Alekh Agarwal, Christoph Dann, Andrea Michi, Marco Gelmi, Yunxuan Li, Raghav Gupta, Avinava Dubey, Alexandre Ramé, Johan Ferret, Geoffrey Cideron, Le Hou, Hongkun Yu, Amr Ahmed, Aranyak Mehta, Léonard Hussenot, Olivier Bachem, Edouard Leurent

Reward-based finetuning is crucial for aligning language policies with intended behaviors (e.g., creativity and safety). A key challenge is to develop steerable language models that trade-off multiple (conflicting) objectives in a flexible and efficient manner. This paper presents Conditional Language Policy (CLP), a general framework for finetuning language models on multiple objectives. Building on techniques from multi-task training and parameter-efficient finetuning, CLP learn steerable models that effectively trade-off conflicting objectives at inference time. Notably, this does not require training or maintaining multiple models to achieve different trade-offs between the objectives. Through extensive experiments and ablations on two summarization datasets, we show that CLP learns steerable language models that outperform and Pareto-dominate the existing approaches for multi-objective finetuning.

comment: 40 pages. Findings of EMNLP 2024

♻ ☆ Utilitarian Algorithm Configuration for Infinite Parameter Spaces

Devon Graham, Kevin Leyton-Brown

Utilitarian algorithm configuration is a general-purpose technique for automatically searching the parameter space of a given algorithm to optimize its performance, as measured by a given utility function, on a given set of inputs. Recently introduced utilitarian configuration procedures offer optimality guarantees about the returned parameterization while provably adapting to the hardness of the underlying problem. However, the applicability of these approaches is severely limited by the fact that they only search a finite, relatively small set of parameters. They cannot effectively search the configuration space of algorithms with continuous or uncountable parameters. In this paper we introduce a new procedure, which we dub COUP (Continuous, Optimistic Utilitarian Procrastination). COUP is designed to search infinite parameter spaces efficiently to find good configurations quickly. Furthermore, COUP maintains the theoretical benefits of previous utilitarian configuration procedures when applied to finite parameter spaces but is significantly faster, both provably and experimentally.

♻ ☆ Safeguard is a Double-edged Sword: Denial-of-service Attack on Large Language Models

Qingzhao Zhang, Ziyang Xiong, Z. Morley Mao

Safety is a paramount concern of large language models (LLMs) in their open deployment. To this end, safeguard methods aim to enforce the ethical and responsible use of LLMs through safety alignment or guardrail mechanisms. However, we found that the malicious attackers could exploit false positives of safeguards, i.e., fooling the safeguard model to block safe content mistakenly, leading to a new denial-of-service (DoS) attack on LLMs. Specifically, by software or phishing attacks on user client software, attackers insert a short, seemingly innocuous adversarial prompt into to user prompt templates in configuration files; thus, this prompt appears in final user requests without visibility in the user interface and is not trivial to identify. By designing an optimization process that utilizes gradient and attention information, our attack can automatically generate seemingly safe adversarial prompts, approximately only 30 characters long, that universally block over 97\% of user requests on Llama Guard 3. The attack presents a new dimension of evaluating LLM safeguards focusing on false positives, fundamentally different from the classic jailbreak.

♻ ☆ Exploring Large Language Models for Feature Selection: A Data-centric Perspective KDD

Dawei Li, Zhen Tan, Huan Liu

The rapid advancement of Large Language Models (LLMs) has significantly influenced various domains, leveraging their exceptional few-shot and zero-shot learning capabilities. In this work, we aim to explore and understand the LLMs-based feature selection methods from a data-centric perspective. We begin by categorizing existing feature selection methods with LLMs into two groups: data-driven feature selection which requires numerical values of samples to do statistical inference and text-based feature selection which utilizes prior knowledge of LLMs to do semantical associations using descriptive context. We conduct experiments in both classification and regression tasks with LLMs in various sizes (e.g., GPT-4, ChatGPT and LLaMA-2). Our findings emphasize the effectiveness and robustness of text-based feature selection methods and showcase their potentials using a real-world medical application. We also discuss the challenges and future opportunities in employing LLMs for feature selection, offering insights for further research and development in this emerging field.

comment: Accepted by SIGKDD Explorations (December 2024)

♻ ☆ STAR: SocioTechnical Approach to Red Teaming Language Models

Laura Weidinger, John Mellor, Bernat Guillen Pegueroles, Nahema Marchal, Ravin Kumar, Kristian Lum, Canfer Akbulut, Mark Diaz, Stevie Bergman, Mikel Rodriguez, Verena Rieser, William Isaac

This research introduces STAR, a sociotechnical framework that improves on current best practices for red teaming safety of large language models. STAR makes two key contributions: it enhances steerability by generating parameterised instructions for human red teamers, leading to improved coverage of the risk surface. Parameterised instructions also provide more detailed insights into model failures at no increased cost. Second, STAR improves signal quality by matching demographics to assess harms for specific groups, resulting in more sensitive annotations. STAR further employs a novel step of arbitration to leverage diverse viewpoints and improve label reliability, treating disagreement not as noise but as a valuable contribution to signal quality.

comment: 8 pages, 5 figures, 5 pages appendix. * denotes equal contribution

♻ ☆ Counter-Current Learning: A Biologically Plausible Dual Network Approach for Deep Learning NeurIPS 2024

Chia-Hsiang Kao, Bharath Hariharan

Despite its widespread use in neural networks, error backpropagation has faced criticism for its lack of biological plausibility, suffering from issues such as the backward locking problem and the weight transport problem. These limitations have motivated researchers to explore more biologically plausible learning algorithms that could potentially shed light on how biological neural systems adapt and learn. Inspired by the counter-current exchange mechanisms observed in biological systems, we propose counter-current learning (CCL), a biologically plausible framework for credit assignment in neural networks. This framework employs a feedforward network to process input data and a feedback network to process targets, with each network enhancing the other through anti-parallel signal propagation. By leveraging the more informative signals from the bottom layer of the feedback network to guide the updates of the top layer of the feedforward network and vice versa, CCL enables the simultaneous transformation of source inputs to target outputs and the dynamic mutual influence of these transformations. Experimental results on MNIST, FashionMNIST, CIFAR10, and CIFAR100 datasets using multi-layer perceptrons and convolutional neural networks demonstrate that CCL achieves comparable performance to other biologically plausible algorithms while offering a more biologically realistic learning mechanism. Furthermore, we showcase the applicability of our approach to an autoencoder task, underscoring its potential for unsupervised representation learning. Our work presents a direction for biologically inspired and plausible learning algorithms, offering an alternative mechanism of learning and adaptation in neural networks.

comment: Accepted at NeurIPS 2024. Code available at https://github.com/IandRover/CCL-NeurIPS24

♻ ☆ Proof of Thought : Neurosymbolic Program Synthesis allows Robust and Interpretable Reasoning NeurIPS 2024

Debargha Ganguly, Srinivasan Iyengar, Vipin Chaudhary, Shivkumar Kalyanaraman

Large Language Models (LLMs) have revolutionized natural language processing, yet they struggle with inconsistent reasoning, particularly in novel domains and complex logical sequences. This research introduces Proof of Thought, a framework that enhances the reliability and transparency of LLM outputs. Our approach bridges LLM-generated ideas with formal logic verification, employing a custom interpreter to convert LLM outputs into First Order Logic constructs for theorem prover scrutiny. Central to our method is an intermediary JSON-based Domain-Specific Language, which by design balances precise logical structures with intuitive human concepts. This hybrid representation enables both rigorous validation and accessible human comprehension of LLM reasoning processes. Key contributions include a robust type system with sort management for enhanced logical integrity, explicit representation of rules for clear distinction between factual and inferential knowledge, and a flexible architecture that allows for easy extension to various domain-specific applications. We demonstrate Proof of Thought's effectiveness through benchmarking on StrategyQA and a novel multimodal reasoning task, showing improved performance in open-ended scenarios. By providing verifiable and interpretable results, our technique addresses critical needs for AI system accountability and sets a foundation for human-in-the-loop oversight in high-stakes domains.

comment: 38th Conference on Neural Information Processing Systems (NeurIPS 2024) System 2 Reasoning At Scale Workshop

♻ ☆ CondTSF: One-line Plugin of Dataset Condensation for Time Series Forecasting NeurIPS 2024

Jianrong Ding, Zhanyu Liu, Guanjie Zheng, Haiming Jin, Linghe Kong

Dataset condensation is a newborn technique that generates a small dataset that can be used in training deep neural networks to lower training costs. The objective of dataset condensation is to ensure that the model trained with the synthetic dataset can perform comparably to the model trained with full datasets. However, existing methods predominantly concentrate on classification tasks, posing challenges in their adaptation to time series forecasting (TS-forecasting). This challenge arises from disparities in the evaluation of synthetic data. In classification, the synthetic data is considered well-distilled if the model trained with the full dataset and the model trained with the synthetic dataset yield identical labels for the same input, regardless of variations in output logits distribution. Conversely, in TS-forecasting, the effectiveness of synthetic data distillation is determined by the distance between predictions of the two models. The synthetic data is deemed well-distilled only when all data points within the predictions are similar. Consequently, TS-forecasting has a more rigorous evaluation methodology compared to classification. To mitigate this gap, we theoretically analyze the optimization objective of dataset condensation for TS-forecasting and propose a new one-line plugin of dataset condensation designated as Dataset Condensation for Time Series Forecasting (CondTSF) based on our analysis. Plugging CondTSF into previous dataset condensation methods facilitates a reduction in the distance between the predictions of the model trained with the full dataset and the model trained with the synthetic dataset, thereby enhancing performance. We conduct extensive experiments on eight commonly used time series datasets. CondTSF consistently improves the performance of all previous dataset condensation methods across all datasets, particularly at low condensing ratios.

comment: Accepted by NeurIPS 2024, the project can be found at https://github.com/RafaDD/CondTSF

♻ ☆ Do Large Language Models Truly Grasp Mathematics? An Empirical Exploration

Wei Xie, Shuoyoucheng Ma, Zhenhua Wang, Enze Wang, Baosheng Wang, Jinshu Su

Despite their proficiency in math tasks, the mechanisms underlying LLMs' mathematical reasoning abilities remain a subject of debate. Recent studies suggest that chain-of-thought (CoT) prompts can bolster mathematical reasoning by encouraging LLMs to employ human-like logical reasoning (System 2), enabling them to excel on the Cognitive Reflection Test (CRT). To assess whether LLMs genuinely possess System 2-like logical reasoning, we introduced targeted modifications to CRT problems. Our findings reveal that, despite the use of CoT prompts, mainstream LLMs, including the latest o1-preview model, continue to exhibit a significant error rate. Further analysis indicates that they predominantly rely on System 1-like intuitive reasoning and pattern matching derived from training data, rather than demonstrating mastery of mathematical thinking. This discovery challenges the prevailing notion that LLMs possess genuine logical reasoning abilities and that CoT can enhance them. Consequently, this work may temper overly optimistic projections regarding LLMs' advancement toward artificial general intelligence.

♻ ☆ StockGPT: A GenAI Model for Stock Prediction and Trading

Dat Mai

This paper introduces StockGPT, an autoregressive ``number'' model trained and tested on 70 million daily U.S.\ stock returns over nearly 100 years. Treating each return series as a sequence of tokens, StockGPT automatically learns the hidden patterns predictive of future returns via its attention mechanism. On a held-out test sample from 2001 to 2023, daily and monthly rebalanced long-short portfolios formed from StockGPT predictions yield strong performance. The StockGPT-based portfolios span momentum and long-/short-term reversals, eliminating the need for manually crafted price-based strategies, and yield highly significant alphas against leading stock market factors, suggesting a novel AI pricing effect. This highlights the immense promise of generative AI in surpassing human in making complex financial investment decisions.

comment: 26 pages, 3 figures, 8 tables

♻ ☆ Certifiably Robust Policies for Uncertain Parametric Environments

Yannik Schnitzer, Alessandro Abate, David Parker

We present a data-driven approach for producing policies that are provably robust across unknown stochastic environments. Existing approaches can learn models of a single environment as an interval Markov decision processes (IMDP) and produce a robust policy with a probably approximately correct (PAC) guarantee on its performance. However these are unable to reason about the impact of environmental parameters underlying the uncertainty. We propose a framework based on parametric Markov decision processes (MDPs) with unknown distributions over parameters. We learn and analyse IMDPs for a set of unknown sample environments induced by parameters. The key challenge is then to produce meaningful performance guarantees that combine the two layers of uncertainty: (1) multiple environments induced by parameters with an unknown distribution; (2) unknown induced environments which are approximated by IMDPs. We present a novel approach based on scenario optimisation that yields a single PAC guarantee quantifying the risk level for which a specified performance level can be assured in unseen environments, plus a means to trade-off risk and performance. We implement and evaluate our framework using multiple robust policy generation methods on a range of benchmarks. We show that our approach produces tight bounds on a policy's performance with high confidence.

♻ ☆ SCA: Highly Efficient Semantic-Consistent Unrestricted Adversarial Attack

Zihao Pan, Weibin Wu, Yuhang Cao, Zibin Zheng

Deep neural network based systems deployed in sensitive environments are vulnerable to adversarial attacks. Unrestricted adversarial attacks typically manipulate the semantic content of an image (e.g., color or texture) to create adversarial examples that are both effective and photorealistic. Recent works have utilized the diffusion inversion process to map images into a latent space, where high-level semantics are manipulated by introducing perturbations. However, they often results in substantial semantic distortions in the denoised output and suffers from low efficiency. In this study, we propose a novel framework called Semantic-Consistent Unrestricted Adversarial Attacks (SCA), which employs an inversion method to extract edit-friendly noise maps and utilizes Multimodal Large Language Model (MLLM) to provide semantic guidance throughout the process. Under the condition of rich semantic information provided by MLLM, we perform the DDPM denoising process of each step using a series of edit-friendly noise maps, and leverage DPM Solver++ to accelerate this process, enabling efficient sampling with semantic consistency. Compared to existing methods, our framework enables the efficient generation of adversarial examples that exhibit minimal discernible semantic changes. Consequently, we for the first time introduce Semantic-Consistent Adversarial Examples (SCAE). Extensive experiments and visualizations have demonstrated the high efficiency of SCA, particularly in being on average 12 times faster than the state-of-the-art attacks. Our research can further draw attention to the security of multimedia information.

♻ ☆ PnLCalib: Sports Field Registration via Points and Lines Optimization

Marc Gutiérrez-Pérez, Antonio Agudo

Camera calibration in broadcast sports videos presents numerous challenges for accurate sports field registration due to multiple camera angles, varying camera parameters, and frequent occlusions of the field. Traditional search-based methods depend on initial camera pose estimates, which can struggle in non-standard positions and dynamic environments. In response, we propose an optimization-based calibration pipeline that leverages a 3D soccer field model and a predefined set of keypoints to overcome these limitations. Our method also introduces a novel refinement module that improves initial calibration by using detected field lines in a non-linear optimization process. This approach outperforms existing techniques in both multi-view and single-view 3D camera calibration tasks, while maintaining competitive performance in homography estimation. Extensive experimentation on real-world soccer datasets, including SoccerNet-Calibration, WorldCup 2014, and TS-WorldCup, highlights the robustness and accuracy of our method across diverse broadcast scenarios. Our approach offers significant improvements in camera calibration precision and reliability.

comment: Extended version of "No Bells, Just Whistles: Sports Field Registration Leveraging Geometric Properties"

♻ ☆ Posterior Sampling-based Online Learning for Episodic POMDPs

Dengwang Tang, Dongze Ye, Rahul Jain, Ashutosh Nayyar, Pierluigi Nuzzo

Learning in POMDPs is known to be significantly harder than in MDPs. In this paper, we consider the online learning problem for episodic POMDPs with unknown transition and observation models. We propose a Posterior Sampling-based reinforcement learning algorithm for POMDPs (PS4POMDPs), which is much simpler and more implementable compared to state-of-the-art optimism-based online learning algorithms for POMDPs. We show that the Bayesian regret of the proposed algorithm scales as the square root of the number of episodes and is polynomial in the other parameters. In a general setting, the regret scales exponentially in the horizon length $H$, and we show that this is inevitable by providing a lower bound. However, when the POMDP is undercomplete and weakly revealing (a common assumption in the recent literature), we establish a polynomial Bayesian regret bound. We finally propose a posterior sampling algorithm for multi-agent POMDPs, and show it too has sublinear regret.

comment: 41 pages, 9 figures

♻ ☆ PixLore: A Dataset-driven Approach to Rich Image Captioning

Diego Bonilla-Salvador, Marcelino Martínez-Sober, Joan Vila-Francés, Antonio José Serrano-López, Pablo Rodríguez-Belenguer, Fernando Mateo

In the domain of vision-language integration, generating detailed image captions poses a significant challenge due to the lack of curated and rich datasets. This study introduces PixLore, a novel method that leverages Querying Transformers through the fine-tuning of the BLIP-2 model using the LoRa method on a standard commercial GPU. The followed approach, which involves training on a carefully assembled dataset from state-of-the-art Computer Vision models combined and augmented by ChatGPT, addresses the question of whether intricate image understanding can be achieved with an ensemble of smaller-scale models, referred to as Knowledge Stitching. Comparative evaluations against major models such as GPT-4 and Google Bard demonstrate that PixLore-2.7B, despite having considerably fewer parameters, is rated higher than the existing State-of-the-Art models in over half of the assessments. Precisely, PixLore outperform Bard and BLIP-2, which score approximately 35.18% and 27.98% lower than PixLore in the task of image captioning. This research not only presents a groundbreaking approach but also highlights the importance of well-curated datasets in enhancing the performance of smaller models.

comment: Paper in preprint pending of publication

♻ ☆ Anomaly Prediction: A Novel Approach with Explicit Delay and Horizon

Jiang You, Arben Cela, René Natowicz, Jacob Ouanounou, Patrick Siarry

Anomaly detection in time series data is a critical challenge across various domains. Traditional methods typically focus on identifying anomalies in immediate subsequent steps, often underestimating the significance of temporal dynamics such as delay time and horizons of anomalies, which generally require extensive post-analysis. This paper introduces a novel approach for time series anomaly prediction, incorporating temporal information directly into the prediction results. We propose a new dataset specifically designed to evaluate this approach and conduct comprehensive experiments using several state-of-the-art methods. Our results demonstrate the efficacy of our approach in providing timely and accurate anomaly predictions, setting a new benchmark for future research in this field.

♻ ☆ Acquiring Better Load Estimates by Combining Anomaly and Change Point Detection in Power Grid Time-series Measurements

Roel Bouman, Linda Schmeitz, Luco Buise, Jacco Heres, Yuliya Shapovalova, Tom Heskes

In this paper we present novel methodology for automatic anomaly and switch event filtering to improve load estimation in power grid systems. By leveraging unsupervised methods with supervised optimization, our approach prioritizes interpretability while ensuring robust and generalizable performance on unseen data. Through experimentation, a combination of binary segmentation for change point detection and statistical process control for anomaly detection emerges as the most effective strategy, specifically when ensembled in a novel sequential manner. Results indicate the clear wasted potential when filtering is not applied. The automatic load estimation is also fairly accurate, with approximately 90% of estimates falling within a 10% error margin, with only a single significant failure in both the minimum and maximum load estimates across 60 measurements in the test set. Our methodology's interpretability makes it particularly suitable for critical infrastructure planning, thereby enhancing decision-making processes.

comment: All code can be found at: https://github.com/RoelBouman/StormPhase2

♻ ☆ Do LLMs Have Distinct and Consistent Personality? TRAIT: Personality Testset designed for LLMs with Psychometrics

Seungbeen Lee, Seungwon Lim, Seungju Han, Giyeong Oh, Hyungjoo Chae, Jiwan Chung, Minju Kim, Beong-woo Kwak, Yeonsoo Lee, Dongha Lee, Jinyoung Yeo, Youngjae Yu

Recent advancements in Large Language Models (LLMs) have led to their adaptation in various domains as conversational agents. We wonder: can personality tests be applied to these agents to analyze their behavior, similar to humans? We introduce TRAIT, a new benchmark consisting of 8K multi-choice questions designed to assess the personality of LLMs. TRAIT is built on two psychometrically validated small human questionnaires, Big Five Inventory (BFI) and Short Dark Triad (SD-3), enhanced with the ATOMIC-10X knowledge graph to a variety of real-world scenarios. TRAIT also outperforms existing personality tests for LLMs in terms of reliability and validity, achieving the highest scores across four key metrics: Content Validity, Internal Validity, Refusal Rate, and Reliability. Using TRAIT, we reveal two notable insights into personalities of LLMs: 1) LLMs exhibit distinct and consistent personality, which is highly influenced by their training data (e.g., data used for alignment tuning), and 2) current prompting techniques have limited effectiveness in eliciting certain traits, such as high psychopathy or low conscientiousness, suggesting the need for further research in this direction.

comment: Preprint; Under review

♻ ☆ Gradient-based Jailbreak Images for Multimodal Fusion Models

Javier Rando, Hannah Korevaar, Erik Brinkman, Ivan Evtimov, Florian Tramèr

Augmenting language models with image inputs may enable more effective jailbreak attacks through continuous optimization, unlike text inputs that require discrete optimization. However, new multimodal fusion models tokenize all input modalities using non-differentiable functions, which hinders straightforward attacks. In this work, we introduce the notion of a tokenizer shortcut that approximates tokenization with a continuous function and enables continuous optimization. We use tokenizer shortcuts to create the first end-to-end gradient image attacks against multimodal fusion models. We evaluate our attacks on Chameleon models and obtain jailbreak images that elicit harmful information for 72.5% of prompts. Jailbreak images outperform text jailbreaks optimized with the same objective and require 3x lower compute budget to optimize 50x more input tokens. Finally, we find that representation engineering defenses, like Circuit Breakers, trained only on text attacks can effectively transfer to adversarial image inputs.

♻ ☆ TargetCall: Eliminating the Wasted Computation in Basecalling via Pre-Basecalling Filtering

Meryem Banu Cavlak, Gagandeep Singh, Mohammed Alser, Can Firtina, Joël Lindegger, Mohammad Sadrosadati, Nika Mansouri Ghiasi, Can Alkan, Onur Mutlu

Basecalling is an essential step in nanopore sequencing analysis where the raw signals of nanopore sequencers are converted into nucleotide sequences, i.e., reads. State-of-the-art basecallers employ complex deep learning models to achieve high basecalling accuracy. This makes basecalling computationally inefficient and memory-hungry, bottlenecking the entire genome analysis pipeline. However, for many applications, the majority of reads do no match the reference genome of interest (i.e., target reference) and thus are discarded in later steps in the genomics pipeline, wasting the basecalling computation. To overcome this issue, we propose TargetCall, the first pre-basecalling filter to eliminate the wasted computation in basecalling. TargetCall's key idea is to discard reads that will not match the target reference (i.e., off-target reads) prior to basecalling. TargetCall consists of two main components: (1) LightCall, a lightweight neural network basecaller that produces noisy reads; and (2) Similarity Check, which labels each of these noisy reads as on-target or off-target by matching them to the target reference. Our thorough experimental evaluations show that TargetCall 1) improves the end-to-end basecalling runtime performance of the state-of-the-art basecaller by 3.31x while maintaining high (98.88%) recall in keeping on-target reads, 2) maintains high accuracy in downstream analysis, and 3) achieves better runtime performance, throughput, recall, precision, and generality compared to prior works. TargetCall is available at https://github.com/CMU-SAFARI/TargetCall.

♻ ☆ Interpreting Context Look-ups in Transformers: Investigating Attention-MLP Interactions EMNLP 2024

Clement Neo, Shay B. Cohen, Fazl Barez

Understanding the inner workings of large language models (LLMs) is crucial for advancing their theoretical foundations and real-world applications. While the attention mechanism and multi-layer perceptrons (MLPs) have been studied independently, their interactions remain largely unexplored. This study investigates how attention heads and next-token neurons interact in LLMs to predict new words. We propose a methodology to identify next-token neurons, find prompts that highly activate them, and determine the upstream attention heads responsible. We then generate and evaluate explanations for the activity of these attention heads in an automated manner. Our findings reveal that some attention heads recognize specific contexts relevant to predicting a token and activate a downstream token-predicting neuron accordingly. This mechanism provides a deeper understanding of how attention heads work with MLP neurons to perform next-token prediction. Our approach offers a foundation for further research into the intricate workings of LLMs and their impact on text generation and understanding.

comment: Accepted to EMNLP 2024 Main Conference

♻ ☆ Do Large Language Models Have an English Accent? Evaluating and Improving the Naturalness of Multilingual LLMs

Yanzhu Guo, Simone Conia, Zelin Zhou, Min Li, Saloni Potdar, Henry Xiao

Current Large Language Models (LLMs) are predominantly designed with English as the primary language, and even the few that are multilingual tend to exhibit strong English-centric biases. Much like speakers who might produce awkward expressions when learning a second language, LLMs often generate unnatural outputs in non-English languages, reflecting English-centric patterns in both vocabulary and grammar. Despite the importance of this issue, the naturalness of multilingual LLM outputs has received limited attention. In this paper, we address this gap by introducing novel automatic corpus-level metrics to assess the lexical and syntactic naturalness of LLM outputs in a multilingual context. Using our new metrics, we evaluate state-of-the-art LLMs on a curated benchmark in French and Chinese, revealing a tendency towards English-influenced patterns. To mitigate this issue, we also propose a simple and effective alignment method to improve the naturalness of an LLM in a target language and domain, achieving consistent improvements in naturalness without compromising the performance on general-purpose benchmarks. Our work highlights the importance of developing multilingual metrics, resources and methods for the new wave of multilingual LLMs.

♻ ☆ ShapefileGPT: A Multi-Agent Large Language Model Framework for Automated Shapefile Processing

Qingming Lin, Rui Hu, Huaxia Li, Sensen Wu, Yadong Li, Kai Fang, Hailin Feng, Zhenhong Du, Liuchang Xu

Vector data is one of the two core data structures in geographic information science (GIS), essential for accurately storing and representing geospatial information. Shapefile, the most widely used vector data format, has become the industry standard supported by all major geographic information systems. However, processing this data typically requires specialized GIS knowledge and skills, creating a barrier for researchers from other fields and impeding interdisciplinary research in spatial data analysis. Moreover, while large language models (LLMs) have made significant advancements in natural language processing and task automation, they still face challenges in handling the complex spatial and topological relationships inherent in GIS vector data. To address these challenges, we propose ShapefileGPT, an innovative framework powered by LLMs, specifically designed to automate Shapefile tasks. ShapefileGPT utilizes a multi-agent architecture, in which the planner agent is responsible for task decomposition and supervision, while the worker agent executes the tasks. We developed a specialized function library for handling Shapefiles and provided comprehensive API documentation, enabling the worker agent to operate Shapefiles efficiently through function calling. For evaluation, we developed a benchmark dataset based on authoritative textbooks, encompassing tasks in categories such as geometric operations and spatial queries. ShapefileGPT achieved a task success rate of 95.24%, outperforming the GPT series models. In comparison to traditional LLMs, ShapefileGPT effectively handles complex vector data analysis tasks, overcoming the limitations of traditional LLMs in spatial analysis. This breakthrough opens new pathways for advancing automation and intelligence in the GIS field, with significant potential in interdisciplinary data analysis and application contexts.

♻ ☆ Enhancing Interaction Modeling with Agent Selection and Physical Coefficient for Trajectory Prediction SP

Shiji Huang, Lei Ye, Min Chen, Wenhai Luo, Dihong Wang, Chenqi Xu, Deyuan Liang

A thorough understanding of the interaction between the target agent and surrounding agents is a prerequisite for accurate trajectory prediction. Although many methods have been explored, they all assign correlation coefficients to surrounding agents in a purely learning-based manner. In this study, we present ASPILin, which manually selects interacting agents and calculates their correlations instead of attention scores. Surprisingly, these simple modifications can significantly improve prediction performance and substantially reduce computational costs. Additionally, ASPILin models the interacting agents at each past time step separately, rather than only modeling the interacting agents at the current time step. This clarifies the causal chain of the target agent's historical trajectory and helps the model better understand dynamic interactions. We intentionally simplified our model in other aspects, such as map encoding. Remarkably, experiments conducted on the INTERACTION, highD, and CitySim datasets demonstrate that our method is efficient and straightforward, outperforming other state-of-the-art methods.

comment: code:https://github.com/kkk00714/ASPILin

♻ ☆ Generative AI Models for Different Steps in Architectural Design: A Literature Review

Chengyuan Li, Tianyu Zhang, Xusheng Du, Ye Zhang, Haoran Xie

Recent advances in generative artificial intelligence (AI) technologies have been significantly driven by models such as generative adversarial networks (GANs), variational autoencoders (VAEs), and denoising diffusion probabilistic models (DDPMs). Although architects recognize the potential of generative AI in design, personal barriers often restrict their access to the latest technological developments, thereby causing the application of generative AI in architectural design to lag behind. Therefore, it is essential to comprehend the principles and advancements of generative AI models and analyze their relevance in architecture applications. This paper first provides an overview of generative AI technologies, with a focus on probabilistic diffusion models (DDPMs), 3D generative models, and foundation models, highlighting their recent developments and main application scenarios. Then, the paper explains how the abovementioned models could be utilized in architecture. We subdivide the architectural design process into six steps and review related research projects in each step from 2020 to the present. Lastly, this paper discusses potential future directions for applying generative AI in the architectural design steps. This research can help architects quickly understand the development and latest progress of generative AI and contribute to the further development of intelligent architecture.

comment: 34 pages, 14 figures, accepted by Frontiers of Architectural Research

♻ ☆ On the limits of agency in agent-based models

Ayush Chopra, Shashank Kumar, Nurullah Giray-Kuru, Ramesh Raskar, Arnau Quera-Bofarull

Agent-based modeling (ABM) seeks to understand the behavior of complex systems by simulating a collection of agents that act and interact within an environment. Their practical utility requires capturing realistic environment dynamics and adaptive agent behavior while efficiently simulating million-size populations. Recent advancements in large language models (LLMs) present an opportunity to enhance ABMs by using LLMs as agents with further potential to capture adaptive behavior. However, the computational infeasibility of using LLMs for large populations has hindered their widespread adoption. In this paper, we introduce AgentTorch -- a framework that scales ABMs to millions of agents while capturing high-resolution agent behavior using LLMs. We benchmark the utility of LLMs as ABM agents, exploring the trade-off between simulation scale and individual agency. Using the COVID-19 pandemic as a case study, we demonstrate how AgentTorch can simulate 8.4 million agents representing New York City, capturing the impact of isolation and employment behavior on health and economic outcomes. We compare the performance of different agent architectures based on heuristic and LLM agents in predicting disease waves and unemployment rates. Furthermore, we showcase AgentTorch's capabilities for retrospective, counterfactual, and prospective analyses, highlighting how adaptive agent behavior can help overcome the limitations of historical data in policy design. AgentTorch is an open-source project actively being used for policy-making and scientific discovery around the world. The framework is available here: github.com/AgentTorch/AgentTorch.

comment: 19 pages, 5 appendices, 5 figures

♻ ☆ A Review of Prominent Paradigms for LLM-Based Agents: Tool Use (Including RAG), Planning, and Feedback Learning

Xinzhe Li

Tool use, planning, and feedback learning are currently three prominent paradigms for developing Large Language Model (LLM)-based agents across various tasks. Although numerous frameworks have been devised for each paradigm, their intricate workflows and inconsistent taxonomy create challenges in understanding and reviewing the frameworks across different paradigms. This survey introduces a unified taxonomy to systematically review and discuss these frameworks. Specifically, 1) the taxonomy defines environments/tasks, common LLM-profiled roles or LMPRs (policy models, evaluators, and dynamic models), and universally applicable workflows found in prior work, and 2) it enables a comparison of key perspectives on the implementations of LMPRs and workflow designs across different agent paradigms and frameworks. 3) Finally, we identify three limitations in existing workflow designs and systematically discuss the future work.

comment: Under Review

♻ ☆ GRAMMAR: Grounded and Modular Methodology for Assessment of Closed-Domain Retrieval-Augmented Language Model

Xinzhe Li, Ming Liu, Shang Gao

Retrieval-Augmented Generation (RAG) systems are widely used across various industries for querying closed-domain and in-house knowledge bases. However, evaluating these systems presents significant challenges due to the private nature of closed-domain data and a scarcity of queries with verifiable ground truths. Moreover, there is a lack of analytical methods to diagnose problematic modules and identify types of failure, such as those caused by knowledge deficits or issues with robustness. To address these challenges, we introduce GRAMMAR (GRounded And Modular Methodology for Assessment of RAG), an evaluation framework comprising a grounded data generation process and an evaluation protocol that effectively pinpoints defective modules. Our validation experiments reveal that GRAMMAR provides a reliable approach for identifying vulnerable modules and supports hypothesis testing for textual form vulnerabilities. An open-source tool accompanying this framework is available in our GitHub repository (see https://github.com/xinzhel/grammar), allowing for easy reproduction of our results and enabling reliable and modular evaluation in closed-domain settings.

comment: Under Review

♻ ☆ Collaborative AI in Sentiment Analysis: System Architecture, Data Prediction and Deployment Strategies

Chaofeng Zhang, Jia Hou, Xueting Tan, Gaolei Li, Caijuan Chen

The advancement of large language model (LLM) based artificial intelligence technologies has been a game-changer, particularly in sentiment analysis. This progress has enabled a shift from highly specialized research environments to practical, widespread applications within the industry. However, integrating diverse AI models for processing complex multimodal data and the associated high costs of feature extraction presents significant challenges. Motivated by the marketing oriented software development +needs, our study introduces a collaborative AI framework designed to efficiently distribute and resolve tasks across various AI systems to address these issues. Initially, we elucidate the key solutions derived from our development process, highlighting the role of generative AI models like \emph{chatgpt}, \emph{google gemini} in simplifying intricate sentiment analysis tasks into manageable, phased objectives. Furthermore, we present a detailed case study utilizing our collaborative AI system in edge and cloud, showcasing its effectiveness in analyzing sentiments across diverse online media channels.

♻ ☆ Bayesian Analysis of Combinatorial Gaussian Process Bandits

Jack Sandberg, Niklas Åkerblom, Morteza Haghir Chehreghani

We consider the combinatorial volatile Gaussian process (GP) semi-bandit problem. Each round, an agent is provided a set of available base arms and must select a subset of them to maximize the long-term cumulative reward. We study the Bayesian setting and provide novel Bayesian cumulative regret bounds for three GP-based algorithms: GP-UCB, GP-BayesUCB and GP-TS. Our bounds extend previous results for GP-UCB and GP-TS to the infinite, volatile and combinatorial setting, and to the best of our knowledge, we provide the first regret bound for GP-BayesUCB. Volatile arms encompass other widely considered bandit problems such as contextual bandits. Furthermore, we employ our framework to address the challenging real-world problem of online energy-efficient navigation, where we demonstrate its effectiveness compared to the alternatives.

comment: 32 pages, 10 figures

♻ ☆ Conformal Prediction for Causal Effects of Continuous Treatments

Maresa Schröder, Dennis Frauen, Jonas Schweisthal, Konstantin Heß, Valentyn Melnychuk, Stefan Feuerriegel

Uncertainty quantification of causal effects is crucial for safety-critical applications such as personalized medicine. A powerful approach for this is conformal prediction, which has several practical benefits due to model-agnostic finite-sample guarantees. Yet, existing methods for conformal prediction of causal effects are limited to binary/discrete treatments and make highly restrictive assumptions such as known propensity scores. In this work, we provide a novel conformal prediction method for potential outcomes of continuous treatments. We account for the additional uncertainty introduced through propensity estimation so that our conformal prediction intervals are valid even if the propensity score is unknown. Our contributions are three-fold: (1) We derive finite-sample prediction intervals for potential outcomes of continuous treatments. (2) We provide an algorithm for calculating the derived intervals. (3) We demonstrate the effectiveness of the conformal prediction intervals in experiments on synthetic and real-world datasets. To the best of our knowledge, we are the first to propose conformal prediction for continuous treatments when the propensity score is unknown and must be estimated from data.

♻ ☆ From Keywords to Structured Summaries: Streamlining Scholarly Information Access ISWC 2024

Mahsa Shamsabadi, Jennifer D'Souza

This paper highlights the growing importance of information retrieval (IR) engines in the scientific community, addressing the inefficiency of traditional keyword-based search engines due to the rising volume of publications. The proposed solution involves structured records, underpinning advanced information technology (IT) tools, including visualization dashboards, to revolutionize how researchers access and filter articles, replacing the traditional text-heavy approach. This vision is exemplified through a proof of concept centered on the "reproductive number estimate of infectious diseases" research theme, using a fine-tuned large language model (LLM) to automate the creation of structured records to populate a backend database that now goes beyond keywords. The result is a next-generation information access system as an IR method accessible at https://orkg.org/usecases/r0-estimates.

comment: 8 pages, 3 figures | Accepted for publication as a poster paper at the International Semantic Web Conference (ISWC 2024)

♻ ☆ OWL2Vec4OA: Tailoring Knowledge Graph Embeddings for Ontology Alignment

Sevinj Teymurova, Ernesto Jiménez-Ruiz, Tillman Weyde, Jiaoyan Chen

Ontology alignment is integral to achieving semantic interoperability as the number of available ontologies covering intersecting domains is increasing. This paper proposes OWL2Vec4OA, an extension of the ontology embedding system OWL2Vec*. While OWL2Vec* has emerged as a powerful technique for ontology embedding, it currently lacks a mechanism to tailor the embedding to the ontology alignment task. OWL2Vec4OA incorporates edge confidence values from seed mappings to guide the random walk strategy. We present the theoretical foundations, implementation details, and experimental evaluation of our proposed extension, demonstrating its potential effectiveness for ontology alignment tasks.

comment: Accepted to the 6th Knowledge Graph and Semantic Web Conference

♻ ☆ Over-the-Air Federated Learning in Cell-Free MIMO with Long-term Power Constraint

Yifan Wang, Cheng Zhang, Yuanndon Zhuang, Mingzeng Dai, Haiming Wang, Yongming Huang

Wireless networks supporting artificial intelligence have gained significant attention, with Over-the-Air Federated Learning emerging as a key application due to its unique transmission and distributed computing characteristics. This paper derives error bounds for Over-the-Air Federated Learning in a Cell-free MIMO system and formulates an optimization problem to minimize optimality gap via joint optimization of power control and beamforming. We introduce the MOP-LOFPC algorithm, which employs Lyapunov optimization to decouple long-term constraints across rounds while requiring only causal channel state information. Experimental results demonstrate that MOP-LOFPC achieves a better and more flexible trade-off between the model's training loss and adherence to long-term power constraints compared to existing baselines.

♻ ☆ Improve Value Estimation of Q Function and Reshape Reward with Monte Carlo Tree Search

Jiamian Li

Reinforcement learning has achieved remarkable success in perfect information games such as Go and Atari, enabling agents to compete at the highest levels against human players. However, research in reinforcement learning for imperfect information games has been relatively limited due to the more complex game structures and randomness. Traditional methods face challenges in training and improving performance in imperfect information games due to issues like inaccurate Q value estimation and reward sparsity. In this paper, we focus on Uno, an imperfect information game, and aim to address these problems by reducing Q value overestimation and reshaping reward function. We propose a novel algorithm that utilizes Monte Carlo Tree Search to average the value estimations in Q function. Even though we choose Double Deep Q Learning as the foundational framework in this paper, our method can be generalized and used in any algorithm which needs Q value estimation, such as the Actor-Critic. Additionally, we employ Monte Carlo Tree Search to reshape the reward structure in the game environment. We compare our algorithm with several traditional methods applied to games such as Double Deep Q Learning, Deep Monte Carlo and Neural Fictitious Self Play, and the experiments demonstrate that our algorithm consistently outperforms these approaches, especially as the number of players in Uno increases, indicating a higher level of difficulty.

♻ ☆ Learning to Plan for Retrieval-Augmented Large Language Models from Knowledge Graphs EMNLP2024

Junjie Wang, Mingyang Chen, Binbin Hu, Dan Yang, Ziqi Liu, Yue Shen, Peng Wei, Zhiqiang Zhang, Jinjie Gu, Jun Zhou, Jeff Z. Pan, Wen Zhang, Huajun Chen

Improving the performance of large language models (LLMs) in complex question-answering (QA) scenarios has always been a research focal point. Recent studies have attempted to enhance LLMs' performance by combining step-wise planning with external retrieval. While effective for advanced models like GPT-3.5, smaller LLMs face challenges in decomposing complex questions, necessitating supervised fine-tuning. Previous work has relied on manual annotation and knowledge distillation from teacher LLMs, which are time-consuming and not accurate enough. In this paper, we introduce a novel framework for enhancing LLMs' planning capabilities by using planning data derived from knowledge graphs (KGs). LLMs fine-tuned with this data have improved planning capabilities, better equipping them to handle complex QA tasks that involve retrieval. Evaluations on multiple datasets, including our newly proposed benchmark, highlight the effectiveness of our framework and the benefits of KG-derived planning data.

comment: EMNLP2024 Findings

♻ ☆ Probabilistic ML Verification via Weighted Model Integration

Paolo Morettin, Andrea Passerini, Roberto Sebastiani

In machine learning (ML) verification, the majority of procedures are non-quantitative and therefore cannot be used for verifying probabilistic models, or be applied in domains where hard guarantees are practically unachievable. The probabilistic formal verification (PFV) of ML models is in its infancy, with the existing approaches limited to specific ML models, properties, or both. This contrasts with standard formal methods techniques, whose successful adoption in real-world scenarios is also due to their support for a wide range of properties and diverse systems. We propose a unifying framework for the PFV of ML systems based on Weighted Model Integration (WMI), a relatively recent formalism for probabilistic inference with algebraic and logical constraints. Crucially, reducing the PFV of ML models to WMI enables the verification of many properties of interest over a wide range of systems, addressing multiple limitations of deterministic verification and ad-hoc algorithms. We substantiate the generality of the approach on prototypical tasks involving the verification of group fairness, monotonicity, robustness to noise, probabilistic local robustness and equivalence among predictors. We characterize the challenges related to the scalability of the approach and, through our WMI-based perspective, we show how successful scaling techniques in the ML verification literature can be generalized beyond their original scope.

♻ ☆ Multi-Excitation Projective Simulation with a Many-Body Physics Inspired Inductive Bias

Philip A. LeMaitre, Marius Krumm, Hans J. Briegel

With the impressive progress of deep learning, applications relying on machine learning are increasingly being integrated into daily life. However, most deep learning models have an opaque, oracle-like nature making it difficult to interpret and understand their decisions. This problem led to the development of the field known as eXplainable Artificial Intelligence (XAI). One method in this field known as Projective Simulation (PS) models a chain-of-thought as a random walk of a particle on a graph with vertices that have concepts attached to them. While this description has various benefits, including the possibility of quantization, it cannot be naturally used to model thoughts that combine several concepts simultaneously. To overcome this limitation, we introduce Multi-Excitation Projective Simulation (mePS), a generalization that considers a chain-of-thought to be a random walk of several particles on a hypergraph. A definition for a dynamic hypergraph is put forward to describe the agent's training history along with applications to AI and hypergraph visualization. An inductive bias inspired by the remarkably successful few-body interaction models used in quantum many-body physics is formalized for our classical mePS framework and employed to tackle the exponential complexity associated with naive implementations of hypergraphs. We prove that our inductive bias reduces the complexity from exponential to polynomial, with the exponent representing the cutoff on how many particles can interact. We numerically apply our method to two toy environments and a more complex scenario modelling the diagnosis of a broken computer. These environments demonstrate the resource savings provided by an appropriate choice of inductive bias, as well as showcasing aspects of interpretability. A quantum model for mePS is also briefly outlined and some future directions for it are discussed.

comment: 26 pages, 8 figures; Code repository at https://github.com/MariusKrumm/ManyBodyMEPS. Reorganized main text for better readability

♻ ☆ Regularizing Hidden States Enables Learning Generalizable Reward Model for LLMs NeurIPS 2024

Rui Yang, Ruomeng Ding, Yong Lin, Huan Zhang, Tong Zhang

Reward models trained on human preference data have been proven to effectively align Large Language Models (LLMs) with human intent within the framework of reinforcement learning from human feedback (RLHF). However, current reward models have limited generalization capabilities to unseen prompts and responses, which can lead to an unexpected phenomenon known as reward over-optimization, resulting in a decline in actual performance due to excessive optimization of rewards. While previous research has advocated for constraining policy optimization, our study introduces a novel approach to enhance the reward model's generalization ability against distribution shifts by regularizing the hidden states. Specifically, we retain the base model's language model head and incorporate a suite of text-generation losses to preserve the hidden states' text-generation capabilities, while concurrently learning a reward head behind the same hidden states. Our experimental results demonstrate that the introduced regularization technique markedly improves the accuracy of learned reward models across a variety of out-of-distribution (OOD) tasks and effectively alleviates the over-optimization issue in RLHF, offering a more reliable and robust preference learning paradigm.

comment: NeurIPS 2024

♻ ☆ Understanding Gradient Boosting Classifier: Training, Prediction, and the Role of $γ_j$

Hung-Hsuan Chen

The Gradient Boosting Classifier (GBC) is a widely used machine learning algorithm for binary classification, which builds decision trees iteratively to minimize prediction errors. This document explains the GBC's training and prediction processes, focusing on the computation of terminal node values $\gamma_j$, which are crucial to optimizing the logistic loss function. We derive $\gamma_j$ through a Taylor series approximation and provide a step-by-step pseudocode for the algorithm's implementation. The guide explains the theory of GBC and its practical application, demonstrating its effectiveness in binary classification tasks. We provide a step-by-step example in the appendix to help readers understand.

♻ ☆ CAT: Contrastive Adapter Training for Personalized Image Generation CVPR

Jae Wan Park, Sang Hyun Park, Jun Young Koh, Junha Lee, Min Song

The emergence of various adapters, including Low-Rank Adaptation (LoRA) applied from the field of natural language processing, has allowed diffusion models to personalize image generation at a low cost. However, due to the various challenges including limited datasets and shortage of regularization and computation resources, adapter training often results in unsatisfactory outcomes, leading to the corruption of the backbone model's prior knowledge. One of the well known phenomena is the loss of diversity in object generation, especially within the same class which leads to generating almost identical objects with minor variations. This poses challenges in generation capabilities. To solve this issue, we present Contrastive Adapter Training (CAT), a simple yet effective strategy to enhance adapter training through the application of CAT loss. Our approach facilitates the preservation of the base model's original knowledge when the model initiates adapters. Furthermore, we introduce the Knowledge Preservation Score (KPS) to evaluate CAT's ability to keep the former information. We qualitatively and quantitatively compare CAT's improvement. Finally, we mention the possibility of CAT in the aspects of multi-concept adapter and optimization.

comment: CVPRW 2024

♻ ☆ GPT-SW3: An Autoregressive Language Model for the Nordic Languages

Ariel Ekgren, Amaru Cuba Gyllensten, Felix Stollenwerk, Joey Öhman, Tim Isbister, Evangelia Gogoulou, Fredrik Carlsson, Alice Heiman, Judit Casademont, Magnus Sahlgren

This paper details the process of developing the first native large generative language model for the Nordic languages, GPT-SW3. We cover all parts of the development process, from data collection and processing, training configuration and instruction finetuning, to evaluation and considerations for release strategies. We hope that this paper can serve as a guide and reference for other researchers that undertake the development of large generative models for smaller languages.

♻ ☆ Solving a Stackelberg Game on Transportation Networks in a Dynamic Crime Scenario: A Mixed Approach on Multi-Layer Networks

Sukanya Samanta, Kei Kimura, Makoto Yokoo

Interdicting a criminal with limited police resources is a challenging task as the criminal changes location over time. The size of the large transportation network further adds to the difficulty of this scenario. To tackle this issue, we consider the concept of a layered graph. At each time stamp, we create a copy of the entire transportation network to track the possible movements of both players, the attacker and the defenders. We consider a Stackelberg game in a dynamic crime scenario where the attacker changes location over time while the defenders attempt to interdict the attacker on his escape route. Given a set of defender strategies, the optimal attacker strategy is determined by applying Dijkstra's algorithm on the layered networks. Here, the attacker aims to minimize while the defenders aim to maximize the probability of interdiction. We develop an approximation algorithm on the layered networks to find near-optimal strategy for defenders. The efficacy of the developed approach is compared with the adopted MILP approach. We compare the results in terms of computational time and solution quality. The quality of the results demonstrates the need for the developed approach, as it effectively solves the complex problem within a short amount of time.

♻ ☆ Richelieu: Self-Evolving LLM-Based Agents for AI Diplomacy

Zhenyu Guan, Xiangyu Kong, Fangwei Zhong, Yizhou Wang

Diplomacy is one of the most sophisticated activities in human society, involving complex interactions among multiple parties that require skills in social reasoning, negotiation, and long-term strategic planning. Previous AI agents have demonstrated their ability to handle multi-step games and large action spaces in multi-agent tasks. However, diplomacy involves a staggering magnitude of decision spaces, especially considering the negotiation stage required. While recent agents based on large language models (LLMs) have shown potential in various applications, they still struggle with extended planning periods in complex multi-agent settings. Leveraging recent technologies for LLM-based agents, we aim to explore AI's potential to create a human-like agent capable of executing comprehensive multi-agent missions by integrating three fundamental capabilities: 1) strategic planning with memory and reflection; 2) goal-oriented negotiation with social reasoning; and 3) augmenting memory through self-play games for self-evolution without human in the loop.

♻ ☆ LVBench: An Extreme Long Video Understanding Benchmark

Weihan Wang, Zehai He, Wenyi Hong, Yean Cheng, Xiaohan Zhang, Ji Qi, Xiaotao Gu, Shiyu Huang, Bin Xu, Yuxiao Dong, Ming Ding, Jie Tang

Recent progress in multimodal large language models has markedly enhanced the understanding of short videos (typically under one minute), and several evaluation datasets have emerged accordingly. However, these advancements fall short of meeting the demands of real-world applications such as embodied intelligence for long-term decision-making, in-depth movie reviews and discussions, and live sports commentary, all of which require comprehension of long videos spanning several hours. To address this gap, we introduce LVBench, a benchmark specifically designed for long video understanding. Our dataset comprises publicly sourced videos and encompasses a diverse set of tasks aimed at long video comprehension and information extraction. LVBench is designed to challenge multimodal models to demonstrate long-term memory and extended comprehension capabilities. Our extensive evaluations reveal that current multimodal models still underperform on these demanding long video understanding tasks. Through LVBench, we aim to spur the development of more advanced models capable of tackling the complexities of long video comprehension. Our data and code are publicly available at: https://lvbench.github.io.

♻ ☆ Generative AI Security: Challenges and Countermeasures

Banghua Zhu, Norman Mu, Jiantao Jiao, David Wagner

Generative AI's expanding footprint across numerous industries has led to both excitement and increased scrutiny. This paper delves into the unique security challenges posed by Generative AI, and outlines potential research directions for managing these risks.

♻ ☆ OpenMU: Your Swiss Army Knife for Music Understanding

Mengjie Zhao, Zhi Zhong, Zhuoyuan Mao, Shiqi Yang, Wei-Hsiang Liao, Shusuke Takahashi, Hiromi Wakaki, Yuki Mitsufuji

We present OpenMU-Bench, a large-scale benchmark suite for addressing the data scarcity issue in training multimodal language models to understand music. To construct OpenMU-Bench, we leveraged existing datasets and bootstrapped new annotations. OpenMU-Bench also broadens the scope of music understanding by including lyrics understanding and music tool usage. Using OpenMU-Bench, we trained our music understanding model, OpenMU, with extensive ablations, demonstrating that OpenMU outperforms baseline models such as MU-Llama. Both OpenMU and OpenMU-Bench are open-sourced to facilitate future research in music understanding and to enhance creative music production efficiency.

comment: Resources: https://github.com/mzhaojp22/openmu

♻ ☆ ConfusedPilot: Confused Deputy Risks in RAG-based LLMs

Ayush RoyChowdhury, Mulong Luo, Prateek Sahu, Sarbartha Banerjee, Mohit Tiwari

Retrieval augmented generation (RAG) is a process where a large language model (LLM) retrieves useful information from a database and then generates the responses. It is becoming popular in enterprise settings for daily business operations. For example, Copilot for Microsoft 365 has accumulated millions of businesses. However, the security implications of adopting such RAG-based systems are unclear. In this paper, we introduce ConfusedPilot, a class of security vulnerabilities of RAG systems that confuse Copilot and cause integrity and confidentiality violations in its responses. First, we investigate a vulnerability that embeds malicious text in the modified prompt in RAG, corrupting the responses generated by the LLM. Second, we demonstrate a vulnerability that leaks secret data, which leverages the caching mechanism during retrieval. Third, we investigate how both vulnerabilities can be exploited to propagate misinformation within the enterprise and ultimately impact its operations, such as sales and manufacturing. We also discuss the root cause of these attacks by investigating the architecture of a RAG-based system. This study highlights the security vulnerabilities in today's RAG-based systems and proposes design guidelines to secure future RAG-based systems.

♻ ☆ Learning to Manipulate Anywhere: A Visual Generalizable Framework For Reinforcement Learning

Zhecheng Yuan, Tianming Wei, Shuiqi Cheng, Gu Zhang, Yuanpei Chen, Huazhe Xu

Can we endow visuomotor robots with generalization capabilities to operate in diverse open-world scenarios? In this paper, we propose \textbf{Maniwhere}, a generalizable framework tailored for visual reinforcement learning, enabling the trained robot policies to generalize across a combination of multiple visual disturbance types. Specifically, we introduce a multi-view representation learning approach fused with Spatial Transformer Network (STN) module to capture shared semantic information and correspondences among different viewpoints. In addition, we employ a curriculum-based randomization and augmentation approach to stabilize the RL training process and strengthen the visual generalization ability. To exhibit the effectiveness of Maniwhere, we meticulously design 8 tasks encompassing articulate objects, bi-manual, and dexterous hand manipulation tasks, demonstrating Maniwhere's strong visual generalization and sim2real transfer abilities across 3 hardware platforms. Our experiments show that Maniwhere significantly outperforms existing state-of-the-art methods. Videos are provided at https://gemcollector.github.io/maniwhere/.

comment: Webpage: https://gemcollector.github.io/maniwhere/

♻ ☆ Deep Learning for Detecting and Early Predicting Chronic Obstructive Pulmonary Disease from Spirogram Time Series

Shuhao Mei, Xin Li, Yuxi Zhou, Jiahao Xu, Yong Zhang, Yuxuan Wan, Shan Cao, Qinghao Zhao, Shijia Geng, Junqing Xie, Shengyong Chen, Shenda Hong

Chronic Obstructive Pulmonary Disease (COPD) is a chronic lung disease that causes airflow obstruction. Current methods can only detect COPD from prominent features in spirogram (Volume-Flow time series) but cannot predict future COPD risk from subtle data patterns. We propose a deep learning-based method, DeepSpiro, for early prediction of future COPD risk. DeepSpiro consists of four key components: SpiroSmoother for stabilizing the Volume-Flow curve, SpiroEncoder for capturing volume evolution through key patches of varying lengths, SpiroExplainer for integrating heterogeneous data and explaining predictions through volume attention, and SpiroPredictor for predicting the disease risk of undiagnosed high-risk patients based on key patch concavity, with prediction horizons of 1, 2, 3, 4, 5 years, or even longer. Evaluated on the UK Biobank dataset, DeepSpiro achieved an AUC of 0.8328 for COPD detection and demonstrated strong predictive performance for future COPD risk (p-value < 0.001). DeepSpiro effectively predicts the long-term progression of the disease.

♻ ☆ Timetable Nodes for Public Transport Network

Andrii Rohovyi, Peter J. Stuckey, Toby Walsh

Faster pathfinding in time-dependent transport networks is an important and challenging problem in navigation systems. There are two main types of transport networks: road networks for car driving and public transport route network. The solutions that work well in road networks, such as Time-dependent Contraction Hierarchies and other graph-based approaches, do not usually apply in transport networks. In transport networks, non-graph solutions such as CSA and RAPTOR show the best results compared to graph-based techniques. In our work, we propose a method that advances graph-based approaches by using different optimization techniques from computational geometry to speed up the search process in transport networks. We apply a new pre-computation step, which we call timetable nodes (TTN). Our inspiration comes from an iterative search problem in computational geometry. We implement two versions of the TTN: one uses a Combined Search Tree (TTN-CST), and the second uses Fractional Cascading (TTN-FC). Both of these approaches decrease the asymptotic complexity of reaching new nodes from $O(k\times \log|C|)$ to $O(k + \log(k) + \log(|C|))$, where $k$ is the number of outgoing edges from a node and $|C|$ is the size of the timetable information (total outgoing edges). Our solution suits any other time-dependent networks and can be integrated into other pathfinding algorithms. Our experiments indicate that this pre-computation significantly enhances the performance on high-density graphs. This study showcases how leveraging computational geometry can enhance pathfinding in transport networks, enabling faster pathfinding in scenarios involving large numbers of outgoing edges.

♻ ☆ RotCAtt-TransUNet++: Novel Deep Neural Network for Sophisticated Cardiac Segmentation

Quoc-Bao Nguyen-Le, Tuan-Hy Le, Anh-Triet Do, Quoc-Huy Trinh

Cardiovascular disease remains a predominant global health concern, responsible for a significant portion of mortality worldwide. Accurate segmentation of cardiac medical imaging data is pivotal in mitigating fatality rates associated with cardiovascular conditions. However, existing state-of-the-art (SOTA) neural networks, including both CNN-based and Transformer-based approaches, exhibit limitations in practical applicability due to their inability to effectively capture inter-slice connections alongside intra-slice information. This deficiency is particularly evident in datasets featuring intricate, long-range details along the z-axis, such as coronary arteries in axial views. Additionally, SOTA methods fail to differentiate non-cardiac components from myocardium in segmentation, leading to the "spraying" phenomenon. To address these challenges, we present RotCAtt-TransUNet++, a novel architecture tailored for robust segmentation of complex cardiac structures. Our approach emphasizes modeling global contexts by aggregating multiscale features with nested skip connections in the encoder. It integrates transformer layers to capture interactions between patches and employs a rotatory attention mechanism to capture connectivity between multiple slices (inter-slice information). Additionally, a channel-wise cross-attention gate guides the fused multi-scale channel-wise information and features from decoder stages to bridge semantic gaps. Experimental results demonstrate that our proposed model outperforms existing SOTA approaches across four cardiac datasets and one abdominal dataset. Importantly, coronary arteries and myocardium are annotated with near-perfect accuracy during inference. An ablation study shows that the rotatory attention mechanism effectively transforms embedded vectorized patches in the semantic dimensional space, enhancing segmentation accuracy.

comment: 11 pages, 11 figures

♻ ☆ Quantformer: from attention to profit with a quantitative transformer trading strategy

Zhaofeng Zhang, Banghao Chen, Shengxin Zhu, Nicolas Langrené

In traditional quantitative trading practice, navigating the complicated and dynamic financial market presents a persistent challenge. Fully capturing various market variables, including long-term information, as well as essential signals that may lead to profit remains a difficult task for learning algorithms. In order to tackle this challenge, this paper introduces quantformer, an enhanced neural network architecture based on transformers, to build investment factors. By transfer learning from sentiment analysis, quantformer not only exploits its original inherent advantages in capturing long-range dependencies and modeling complex data relationships, but is also able to solve tasks with numerical inputs and accurately forecast future returns over a given period. This work collects more than 5,000,000 rolling data of 4,601 stocks in the Chinese capital market from 2010 to 2019. The results of this study demonstrated the model's superior performance in predicting stock trends compared with other 100 factor-based quantitative strategies. Notably, the model's innovative use of transformer-liked model to establish factors, in conjunction with market sentiment information, has been shown to enhance the accuracy of trading signals significantly, thereby offering promising implications for the future of quantitative trading strategies.

♻ ☆ Set-based Meta-Interpolation for Few-Task Meta-Learning

Seanie Lee, Bruno Andreis, Kenji Kawaguchi, Juho Lee, Sung Ju Hwang

Meta-learning approaches enable machine learning systems to adapt to new tasks given few examples by leveraging knowledge from related tasks. However, a large number of meta-training tasks are still required for generalization to unseen tasks during meta-testing, which introduces a critical bottleneck for real-world problems that come with only few tasks, due to various reasons including the difficulty and cost of constructing tasks. Recently, several task augmentation methods have been proposed to tackle this issue using domain-specific knowledge to design augmentation techniques to densify the meta-training task distribution. However, such reliance on domain-specific knowledge renders these methods inapplicable to other domains. While Manifold Mixup based task augmentation methods are domain-agnostic, we empirically find them ineffective on non-image domains. To tackle these limitations, we propose a novel domain-agnostic task augmentation method, Meta-Interpolation, which utilizes expressive neural set functions to densify the meta-training task distribution using bilevel optimization. We empirically validate the efficacy of Meta-Interpolation on eight datasets spanning across various domains such as image classification, molecule property prediction, text classification and speech recognition. Experimentally, we show that Meta-Interpolation consistently outperforms all the relevant baselines. Theoretically, we prove that task interpolation with the set function regularizes the meta-learner to improve generalization.

comment: First two authors contributed equally. Name order decided by a coin toss

♻ ☆ RealignDiff: Boosting Text-to-Image Diffusion Model with Coarse-to-fine Semantic Re-alignment

Guian Fang, Zutao Jiang, Jianhua Han, Guansong Lu, Hang Xu, Shengcai Liao, Xiaojun Chang, Xiaodan Liang

Recent advances in text-to-image diffusion models have achieved remarkable success in generating high-quality, realistic images from textual descriptions. However, these approaches have faced challenges in precisely aligning the generated visual content with the textual concepts described in the prompts. In this paper, we propose a two-stage coarse-to-fine semantic re-alignment method, named RealignDiff, aimed at improving the alignment between text and images in text-to-image diffusion models. In the coarse semantic re-alignment phase, a novel caption reward, leveraging the BLIP-2 model, is proposed to evaluate the semantic discrepancy between the generated image caption and the given text prompt. Subsequently, the fine semantic re-alignment stage employs a local dense caption generation module and a re-weighting attention modulation module to refine the previously generated images from a local semantic view. Experimental results on the MS-COCO and ViLG-300 datasets demonstrate that the proposed two-stage coarse-to-fine semantic re-alignment method outperforms other baseline re-alignment techniques by a substantial margin in both visual quality and semantic similarity with the input prompt.

♻ ☆ Quantifying the Gain in Weak-to-Strong Generalization NeurIPS 2024

Moses Charikar, Chirag Pabbaraju, Kirankumar Shiragur

Recent advances in large language models have shown capabilities that are extraordinary and near-superhuman. These models operate with such complexity that reliably evaluating and aligning them proves challenging for humans. This leads to the natural question: can guidance from weak models (like humans) adequately direct the capabilities of strong models? In a recent and somewhat surprising work, Burns et al. (2023) empirically demonstrated that when strong models (like GPT-4) are finetuned using labels generated by weak supervisors (like GPT-2), the strong models outperform their weaker counterparts -- a phenomenon they term weak-to-strong generalization. In this work, we present a theoretical framework for understanding weak-to-strong generalization. Specifically, we show that the improvement in performance achieved by strong models over their weaker counterparts is quantified by the misfit error incurred by the strong model on labels generated by the weaker model. Our theory reveals several curious algorithmic insights. For instance, we can predict the amount by which the strong model will improve over the weak model, and also choose among different weak models to train the strong model, based on its misfit error. We validate our theoretical findings through various empirical assessments.

comment: 19 pages; NeurIPS 2024 camera-ready version with additional experiments, references and discussion

♻ ☆ Real-World Robot Applications of Foundation Models: A Review

Kento Kawaharazuka, Tatsuya Matsushima, Andrew Gambardella, Jiaxian Guo, Chris Paxton, Andy Zeng

Recent developments in foundation models, like Large Language Models (LLMs) and Vision-Language Models (VLMs), trained on extensive data, facilitate flexible application across different tasks and modalities. Their impact spans various fields, including healthcare, education, and robotics. This paper provides an overview of the practical application of foundation models in real-world robotics, with a primary emphasis on the replacement of specific components within existing robot systems. The summary encompasses the perspective of input-output relationships in foundation models, as well as their role in perception, motion planning, and control within the field of robotics. This paper concludes with a discussion of future challenges and implications for practical robot applications.

♻ ☆ TSDS: Data Selection for Task-Specific Model Finetuning

Zifan Liu, Amin Karbasi, Theodoros Rekatsinas

Finetuning foundation models for specific tasks is an emerging paradigm in modern machine learning. The efficacy of task-specific finetuning largely depends on the selection of appropriate training data. We present TSDS (Task-Specific Data Selection), a framework to select data for task-specific model finetuning, guided by a small but representative set of examples from the target task. To do so, we formulate data selection for task-specific finetuning as an optimization problem with a distribution alignment loss based on optimal transport to capture the discrepancy between the selected data and the target distribution. In addition, we add a regularizer to encourage the diversity of the selected data and incorporate kernel density estimation into the regularizer to reduce the negative effects of near-duplicates among the candidate data. We connect our optimization problem to nearest neighbor search and design efficient algorithms to compute the optimal solution based on approximate nearest neighbor search techniques. We evaluate our method on data selection for both continued pretraining and instruction tuning of language models. We show that instruction tuning using data selected by our method with a 1% selection ratio often outperforms using the full dataset and beats the baseline selection methods by 1.5 points in F1 score on average.

comment: 31 pages, 1 figure

♻ ☆ Can visual language models resolve textual ambiguity with visual cues? Let visual puns tell you! EMNLP 2024

Jiwan Chung, Seungwon Lim, Jaehyun Jeon, Seungbeen Lee, Youngjae Yu

Humans possess multimodal literacy, allowing them to actively integrate information from various modalities to form reasoning. Faced with challenges like lexical ambiguity in text, we supplement this with other modalities, such as thumbnail images or textbook illustrations. Is it possible for machines to achieve a similar multimodal understanding capability? In response, we present Understanding Pun with Image Explanations (UNPIE), a novel benchmark designed to assess the impact of multimodal inputs in resolving lexical ambiguities. Puns serve as the ideal subject for this evaluation due to their intrinsic ambiguity. Our dataset includes 1,000 puns, each accompanied by an image that explains both meanings. We pose three multimodal challenges with the annotations to assess different aspects of multimodal literacy; Pun Grounding, Disambiguation, and Reconstruction. The results indicate that various Socratic Models and Visual-Language Models improve over the text-only models when given visual context, particularly as the complexity of the tasks increases.

comment: Accepted as main paper in EMNLP 2024

♻ ☆ RegExplainer: Generating Explanations for Graph Neural Networks in Regression Task NeurIPS 2024

Jiaxing Zhang, Zhuomin Chen, Hao Mei, Dongsheng Luo, Hua Wei

Graph regression is a fundamental task and has received increasing attention in a wide range of graph learning tasks. However, the inference process is often not interpretable. Most existing explanation techniques are limited to understanding GNN behaviors in classification tasks. In this work, we seek an explanation to interpret the graph regression models (XAIG-R). We show that existing methods overlook the distribution shifting and continuously ordered decision boundary, which hinders them away from being applied in the regression tasks. To address these challenges, we propose a novel objective based on the information bottleneck theory and introduce a new mix-up framework, which could support various GNNs in a model-agnostic manner. We further present a contrastive learning strategy to tackle the continuously ordered labels in regression task. To empirically verify the effectiveness of the proposed method, we introduce three benchmark datasets and a real-life dataset for evaluation. Extensive experiments show the effectiveness of the proposed method in interpreting GNN models in regression tasks.

comment: Accepted by NeurIPS 2024

♻ ☆ CV-VAE: A Compatible Video VAE for Latent Generative Video Models

Sijie Zhao, Yong Zhang, Xiaodong Cun, Shaoshu Yang, Muyao Niu, Xiaoyu Li, Wenbo Hu, Ying Shan

Spatio-temporal compression of videos, utilizing networks such as Variational Autoencoders (VAE), plays a crucial role in OpenAI's SORA and numerous other video generative models. For instance, many LLM-like video models learn the distribution of discrete tokens derived from 3D VAEs within the VQVAE framework, while most diffusion-based video models capture the distribution of continuous latent extracted by 2D VAEs without quantization. The temporal compression is simply realized by uniform frame sampling which results in unsmooth motion between consecutive frames. Currently, there lacks of a commonly used continuous video (3D) VAE for latent diffusion-based video models in the research community. Moreover, since current diffusion-based approaches are often implemented using pre-trained text-to-image (T2I) models, directly training a video VAE without considering the compatibility with existing T2I models will result in a latent space gap between them, which will take huge computational resources for training to bridge the gap even with the T2I models as initialization. To address this issue, we propose a method for training a video VAE of latent video models, namely CV-VAE, whose latent space is compatible with that of a given image VAE, e.g., image VAE of Stable Diffusion (SD). The compatibility is achieved by the proposed novel latent space regularization, which involves formulating a regularization loss using the image VAE. Benefiting from the latent space compatibility, video models can be trained seamlessly from pre-trained T2I or video models in a truly spatio-temporally compressed latent space, rather than simply sampling video frames at equal intervals. With our CV-VAE, existing video models can generate four times more frames with minimal finetuning. Extensive experiments are conducted to demonstrate the effectiveness of the proposed video VAE.

comment: Project Page: https://ailab-cvc.github.io/cvvae/index.html

♻ ☆ Do LLMs Have Political Correctness? Analyzing Ethical Biases and Jailbreak Vulnerabilities in AI Systems

Isack Lee, Haebin Seong

Although large language models (LLMs) demonstrate impressive proficiency in various tasks, they present potential safety risks, such as `jailbreaks', where malicious inputs can coerce LLMs into generating harmful content. To address these issues, many LLM developers have implemented various safety measures to align these models. This alignment involves several techniques, including data filtering during pre-training, supervised fine-tuning, reinforcement learning from human feedback, and red-teaming exercises. These methods often introduce deliberate and intentional biases similar to Political Correctness (PC) to ensure the ethical behavior of LLMs. In this paper, we delve into the intentional biases injected into LLMs for safety purposes and examine methods to circumvent these safety alignment techniques. Notably, these intentional biases result in a jailbreaking success rate in GPT-4o models that differs by 20% between non-binary and cisgender keywords and by 16% between white and black keywords, even when the other parts of the prompts are identical. We introduce the concept of PCJailbreak, highlighting the inherent risks posed by these safety-induced biases. Additionally, we propose an efficient defense method PCDefense, which prevents jailbreak attempts by injecting defense prompts prior to generation. PCDefense stands as an appealing alternative to Guard Models, such as Llama-Guard, that require additional inference cost after text generation. Our findings emphasize the urgent need for LLM developers to adopt a more responsible approach when designing and implementing safety measures.

♻ ☆ Context-Aware Assistant Selection for Improved Inference Acceleration with Large Language Models EMNLP

Jerry Huang, Prasanna Parthasarathi, Mehdi Rezagholizadeh, Sarath Chandar

Despite their widespread adoption, large language models (LLMs) remain prohibitive to use under resource constraints, with their ever growing sizes only increasing the barrier for use. One noted issue is the high latency associated with auto-regressive generation, rendering large LLMs use dependent on advanced computing infrastructure. Assisted decoding, where a smaller draft model guides a larger target model's generation, has helped alleviate this, but remains dependent on alignment between the two models. Thus if the draft model is insufficiently capable on some domain relative to the target model, performance can degrade. Alternatively, one can leverage multiple draft models to better cover the expertise of the target, but when multiple black-box draft models are available, selecting an assistant without details about its construction can be difficult. To better understand this decision making problem, we observe it as a contextual bandit, where a policy must choose a draft model based on a context. We show that even without prior knowledge of the draft models, creating an offline dataset from only outputs of independent draft/target models and training a policy over the alignment of these outputs can accelerate performance on multiple domains provided the candidates are effective. Further results show this to hold on various settings with multiple assisted decoding candidates, highlighting its flexibility and the advantageous role that such decision making can play.

comment: 2024 Conference on Empirical Methods in Natural Language Processing (EMNLP); 14 pages (9 pages main content + references + appendix)

♻ ☆ On Catastrophic Inheritance of Large Foundation Models

Hao Chen, Bhiksha Raj, Xing Xie, Jindong Wang

Large foundation models (LFMs) are claiming incredible performances. Yet great concerns have been raised about their mythic and uninterpreted potentials not only in machine learning, but also in various other disciplines. In this position paper, we propose to identify a neglected issue deeply rooted in LFMs: Catastrophic Inheritance, describing the weaknesses and limitations inherited from biased large-scale pre-training data to behaviors of LFMs on the downstream tasks, including samples that are corrupted, long-tailed, noisy, out-of-distributed, to name a few. Such inheritance can potentially cause catastrophes to downstream applications, such as bias, lack of generalization, deteriorated performance, security vulnerability, privacy leakage, and value misalignment. We discuss the challenges behind this issue and propose UIM, a framework to Understand the catastrophic inheritance of LFMs from both pre-training and downstream adaptation, Interpret the implications of catastrophic inheritance on downstream tasks, and how to Mitigate it. UIM aims to unite both the machine learning and social sciences communities for more responsible and promising AI development and deployment.

comment: Accepted by DMLR

♻ ☆ When "Competency" in Reasoning Opens the Door to Vulnerability: Jailbreaking LLMs via Novel Complex Ciphers

Divij Handa, Zehua Zhang, Amir Saeidi, Chitta Baral

Recent advancements in the safety of Large Language Models (LLMs) have primarily focused on mitigating attacks crafted in natural language or in common encryption techniques like Base64. However, new models which often possess better reasoning capabilities, open the door to new attack vectors that were previously non-existent in older models. This seems counter-intuitive at first glance, but these advanced models can decipher more complex cryptic queries that previous models could not, making them susceptible to attacks using such prompts. To exploit this vulnerability, we propose Attacks using Custom Encryptions (ACE), a novel method to jailbreak LLMs by leveraging custom encryption schemes. We evaluate the effectiveness of ACE on four state-of-the-art LLMs, achieving Attack Success Rates (ASR) of up to 66% on close-source models and 88% on open-source models. Building upon this, we introduce Layered Attacks using Custom Encryptions (LACE), which employs multiple layers of encryption through our custom ciphers to further enhance the ASR. Our findings demonstrate that LACE significantly enhances the ability to jailbreak LLMs, increasing the ASR of GPT-4o from 40% to 78%, a 38% improvement. Our results highlight that the advanced capabilities of LLMs introduce unforeseen vulnerabilities to complex attacks. Specifically complex and layered ciphers increase the chance of jailbreaking.

comment: 14 pages, 7 figures

☆ Prioritized Generative Replay

Renhao Wang, Kevin Frans, Pieter Abbeel, Sergey Levine, Alexei A. Efros

Sample-efficient online reinforcement learning often uses replay buffers to store experience for reuse when updating the value function. However, uniform replay is inefficient, since certain classes of transitions can be more relevant to learning. While prioritization of more useful samples is helpful, this strategy can also lead to overfitting, as useful samples are likely to be more rare. In this work, we instead propose a prioritized, parametric version of an agent's memory, using generative models to capture online experience. This paradigm enables (1) densification of past experience, with new generations that benefit from the generative model's generalization capacity and (2) guidance via a family of "relevance functions" that push these generations towards more useful parts of an agent's acquired history. We show this recipe can be instantiated using conditional diffusion models and simple relevance functions such as curiosity- or value-based metrics. Our approach consistently improves performance and sample efficiency in both state- and pixel-based domains. We expose the mechanisms underlying these gains, showing how guidance promotes diversity in our generated transitions and reduces overfitting. We also showcase how our approach can train policies with even higher update-to-data ratios than before, opening up avenues to better scale online RL agents.

☆ ALTA: Compiler-Based Analysis of Transformers

Peter Shaw, James Cohan, Jacob Eisenstein, Kenton Lee, Jonathan Berant, Kristina Toutanova

We propose a new programming language called ALTA and a compiler that can map ALTA programs to Transformer weights. ALTA is inspired by RASP, a language proposed by Weiss et al. (2021), and Tracr (Lindner et al., 2023), a compiler from RASP programs to Transformer weights. ALTA complements and extends this prior work, offering the ability to express loops and to compile programs to Universal Transformers, among other advantages. ALTA allows us to constructively show how Transformers can represent length-invariant algorithms for computing parity and addition, as well as a solution to the SCAN benchmark of compositional generalization tasks, without requiring intermediate scratchpad decoding steps. We also propose tools to analyze cases where the expressibility of an algorithm is established, but end-to-end training on a given training set fails to induce behavior consistent with the desired algorithm. To this end, we explore training from ALTA execution traces as a more fine-grained supervision signal. This enables additional experiments and theoretical analyses relating the learnability of various algorithms to data availability and modeling decisions, such as positional encodings. We make the ALTA framework -- language specification, symbolic interpreter, and weight compiler -- available to the community to enable further applications and insights.

☆ Leveraging Skills from Unlabeled Prior Data for Efficient Online Exploration

Max Wilcoxson, Qiyang Li, Kevin Frans, Sergey Levine

Unsupervised pretraining has been transformative in many supervised domains. However, applying such ideas to reinforcement learning (RL) presents a unique challenge in that fine-tuning does not involve mimicking task-specific data, but rather exploring and locating the solution through iterative self-improvement. In this work, we study how unlabeled prior trajectory data can be leveraged to learn efficient exploration strategies. While prior data can be used to pretrain a set of low-level skills, or as additional off-policy data for online RL, it has been unclear how to combine these ideas effectively for online exploration. Our method SUPE (Skills from Unlabeled Prior data for Exploration) demonstrates that a careful combination of these ideas compounds their benefits. Our method first extracts low-level skills using a variational autoencoder (VAE), and then pseudo-relabels unlabeled trajectories using an optimistic reward model, transforming prior data into high-level, task-relevant examples. Finally, SUPE uses these transformed examples as additional off-policy data for online RL to learn a high-level policy that composes pretrained low-level skills to explore efficiently. We empirically show that SUPE reliably outperforms prior strategies, successfully solving a suite of long-horizon, sparse-reward tasks. Code: https://github.com/rail-berkeley/supe.

comment: 23 pages, 10 figures

☆ ProFL: Performative Robust Optimal Federated Learning

Xue Zheng, Tian Xie, Xuwei Tan, Aylin Yener, Xueru Zhang, Ali Payani, Myungjin Lee

Performative prediction (PP) is a framework that captures distribution shifts that occur during the training of machine learning models due to their deployment. As the trained model is used, its generated data could cause the model to evolve, leading to deviations from the original data distribution. The impact of such model-induced distribution shifts in the federated learning (FL) setup remains unexplored despite being increasingly likely to transpire in real-life use cases. Although Jin et al. (2024) recently extended PP to FL in a straightforward manner, the resulting model only converges to a performative stable point, which may be far from optimal. The methods in Izzo et al. (2021); Miller et al. (2021) can find a performative optimal point in centralized settings, but they require the performative risk to be convex and the training data to be noiseless, assumptions often violated in realistic FL systems. This paper overcomes all of these shortcomings and proposes Performative robust optimal Federated Learning (ProFL), an algorithm that finds performative optimal points in FL from noisy and contaminated data. We present the convergence analysis under the Polyak-Lojasiewicz condition, which applies to non-convex objectives. Extensive experiments on multiple datasets validate our proposed algorithms' efficiency.

comment: 27 pages with Appendix, 18 figures. The paper has been submitted and is currently under review

☆ UnCLe: Unsupervised Continual Learning of Depth Completion

Suchisrit Gangopadhyay, Xien Chen, Michael Chu, Patrick Rim, Hyoungseob Park, Alex Wong

We propose UnCLe, a standardized benchmark for Unsupervised Continual Learning of a multimodal depth estimation task: Depth completion aims to infer a dense depth map from a pair of synchronized RGB image and sparse depth map. We benchmark depth completion models under the practical scenario of unsupervised learning over continuous streams of data. Existing methods are typically trained on a static, or stationary, dataset. However, when adapting to novel non-stationary distributions, they "catastrophically forget" previously learned information. UnCLe simulates these non-stationary distributions by adapting depth completion models to sequences of datasets containing diverse scenes captured from distinct domains using different visual and range sensors. We adopt representative methods from continual learning paradigms and translate them to enable unsupervised continual learning of depth completion. We benchmark these models for indoor and outdoor and investigate the degree of catastrophic forgetting through standard quantitative metrics. Furthermore, we introduce model inversion quality as an additional measure of forgetting. We find that unsupervised continual learning of depth completion is an open problem, and we invite researchers to leverage UnCLe as a development platform.

comment: Preprint

☆ Training Free Guided Flow Matching with Optimal Control

Luran Wang, Chaoran Cheng, Yizhen Liao, Yanru Qu, Ge Liu

Controlled generation with pre-trained Diffusion and Flow Matching models has vast applications. One strategy for guiding ODE-based generative models is through optimizing a target loss $R(x_1)$ while staying close to the prior distribution. Along this line, some recent work showed the effectiveness of guiding flow model by differentiating through its ODE sampling process. Despite the superior performance, the theoretical understanding of this line of methods is still preliminary, leaving space for algorithm improvement. Moreover, existing methods predominately focus on Euclidean data manifold, and there is a compelling need for guided flow methods on complex geometries such as SO(3), which prevails in high-stake scientific applications like protein design. We present OC-Flow, a general and theoretically grounded training-free framework for guided flow matching using optimal control. Building upon advances in optimal control theory, we develop effective and practical algorithms for solving optimal control in guided ODE-based generation and provide a systematic theoretical analysis of the convergence guarantee in both Euclidean and SO(3). We show that existing backprop-through-ODE methods can be interpreted as special cases of Euclidean OC-Flow. OC-Flow achieved superior performance in extensive experiments on text-guided image manipulation, conditional molecule generation, and all-atom peptide design.

☆ Beyond position: how rotary embeddings shape representations and memory in autoregressive transfomers

Valeria Ruscio, Fabrizio Silvestri

Rotary Positional Embeddings (RoPE) enhance positional encoding in Transformer models, yet their full impact on model dynamics remains underexplored. This paper studies how RoPE introduces position-dependent rotations, causing phase shifts in token embeddings that influence higher-frequency components within the model's internal representations. Through spectral analysis, we demonstrate that RoPE's rotation matrices induce oscillatory behaviors in embeddings, affecting information retention across layers and shaping temporal modeling capabilities. We show that activation functions in feed-forward networks interact with RoPE-modulated embeddings to generate harmonics, leading to constructive or destructive interference based on phase alignment. Our findings reveal that phase alignment amplifies activations and sharpens attention, while misalignment weakens activations and disrupts focus on positional patterns. This study underscores the importance of frequency components as intrinsic elements of model behavior, offering new insights beyond traditional analyses.

☆ The Double-Edged Sword of Behavioral Responses in Strategic Classification: Theory and User Studies

Raman Ebrahimi, Kristen Vaccaro, Parinaz Naghizadeh

When humans are subject to an algorithmic decision system, they can strategically adjust their behavior accordingly (``game'' the system). While a growing line of literature on strategic classification has used game-theoretic modeling to understand and mitigate such gaming, these existing works consider standard models of fully rational agents. In this paper, we propose a strategic classification model that considers behavioral biases in human responses to algorithms. We show how misperceptions of a classifier (specifically, of its feature weights) can lead to different types of discrepancies between biased and rational agents' responses, and identify when behavioral agents over- or under-invest in different features. We also show that strategic agents with behavioral biases can benefit or (perhaps, unexpectedly) harm the firm compared to fully rational strategic agents. We complement our analytical results with user studies, which support our hypothesis of behavioral biases in human responses to the algorithm. Together, our findings highlight the need to account for human (cognitive) biases when designing AI systems, and providing explanations of them, to strategic human in the loop.

☆ SPIRE: Synergistic Planning, Imitation, and Reinforcement Learning for Long-Horizon Manipulation

Zihan Zhou, Animesh Garg, Dieter Fox, Caelan Garrett, Ajay Mandlekar

Robot learning has proven to be a general and effective technique for programming manipulators. Imitation learning is able to teach robots solely from human demonstrations but is bottlenecked by the capabilities of the demonstrations. Reinforcement learning uses exploration to discover better behaviors; however, the space of possible improvements can be too large to start from scratch. And for both techniques, the learning difficulty increases proportional to the length of the manipulation task. Accounting for this, we propose SPIRE, a system that first uses Task and Motion Planning (TAMP) to decompose tasks into smaller learning subproblems and second combines imitation and reinforcement learning to maximize their strengths. We develop novel strategies to train learning agents when deployed in the context of a planning system. We evaluate SPIRE on a suite of long-horizon and contact-rich robot manipulation problems. We find that SPIRE outperforms prior approaches that integrate imitation learning, reinforcement learning, and planning by 35% to 50% in average task performance, is 6 times more data efficient in the number of human demonstrations needed to train proficient agents, and learns to complete tasks nearly twice as efficiently. View https://sites.google.com/view/spire-corl-2024 for more details.

comment: Conference on Robot Learning (CoRL) 2024

☆ POD-Attention: Unlocking Full Prefill-Decode Overlap for Faster LLM Inference

Aditya K Kamath, Ramya Prabhu, Jayashree Mohan, Simon Peter, Ramachandran Ramjee, Ashish Panwar

Each request in LLM inference goes through two phases: compute-bound prefill and memory-bandwidth-bound decode. To improve GPU utilization, recent systems use hybrid batching that combines the prefill and decode phases of different requests into the same batch. Hybrid batching works well for linear operations as it amortizes the cost of loading model weights from HBM. However, attention computation in hybrid batches remains inefficient because existing attention kernels are optimized for either prefill or decode. In this paper, we present POD-Attention -- the first GPU kernel that efficiently computes attention for hybrid batches. POD-Attention aims to maximize the utilization of both compute and memory bandwidth by carefully allocating the GPU's resources such that prefill and decode operations happen concurrently on the same multiprocessor. We integrate POD-Attention in a state-of-the-art LLM inference scheduler Sarathi-Serve. POD-Attention speeds up attention computation by up to 75% (mean 28%) and increases LLM serving throughput by up to 22% in offline inference. In online inference, POD-Attention enables lower time-to-first-token (TTFT), time-between-tokens (TBT), and request execution latency versus Sarathi-Serve.

☆ Inferring stability properties of chaotic systems on autoencoders' latent spaces

Elise Özalp, Luca Magri

The data-driven learning of solutions of partial differential equations can be based on a divide-and-conquer strategy. First, the high dimensional data is compressed to a latent space with an autoencoder; and, second, the temporal dynamics are inferred on the latent space with a form of recurrent neural network. In chaotic systems and turbulence, convolutional autoencoders and echo state networks (CAE-ESN) successfully forecast the dynamics, but little is known about whether the stability properties can also be inferred. We show that the CAE-ESN model infers the invariant stability properties and the geometry of the tangent space in the low-dimensional manifold (i.e. the latent space) through Lyapunov exponents and covariant Lyapunov vectors. This work opens up new opportunities for inferring the stability of high-dimensional chaotic systems in latent spaces.

☆ Estimating the Spectral Moments of the Kernel Integral Operator from Finite Sample Matrices

Chanwoo Chun, SueYeon Chung, Daniel D. Lee

Analyzing the structure of sampled features from an input data distribution is challenging when constrained by limited measurements in both the number of inputs and features. Traditional approaches often rely on the eigenvalue spectrum of the sample covariance matrix derived from finite measurement matrices; however, these spectra are sensitive to the size of the measurement matrix, leading to biased insights. In this paper, we introduce a novel algorithm that provides unbiased estimates of the spectral moments of the kernel integral operator in the limit of infinite inputs and features from finitely sampled measurement matrices. Our method, based upon dynamic programming, is efficient and capable of estimating the moments of the operator spectrum. We demonstrate the accuracy of our estimator on radial basis function (RBF) kernels, highlighting its consistency with the theoretical spectra. Furthermore, we showcase the practical utility and robustness of our method in understanding the geometry of learned representations in neural networks.

☆ Federated Transformer: Multi-Party Vertical Federated Learning on Practical Fuzzily Linked Data

Zhaomin Wu, Junyi Hou, Yiqun Diao, Bingsheng He

Federated Learning (FL) is an evolving paradigm that enables multiple parties to collaboratively train models without sharing raw data. Among its variants, Vertical Federated Learning (VFL) is particularly relevant in real-world, cross-organizational collaborations, where distinct features of a shared instance group are contributed by different parties. In these scenarios, parties are often linked using fuzzy identifiers, leading to a common practice termed as multi-party fuzzy VFL. Existing models generally address either multi-party VFL or fuzzy VFL between two parties. Extending these models to practical multi-party fuzzy VFL typically results in significant performance degradation and increased costs for maintaining privacy. To overcome these limitations, we introduce the Federated Transformer (FeT), a novel framework that supports multi-party VFL with fuzzy identifiers. FeT innovatively encodes these identifiers into data representations and employs a transformer architecture distributed across different parties, incorporating three new techniques to enhance performance. Furthermore, we have developed a multi-party privacy framework for VFL that integrates differential privacy with secure multi-party computation, effectively protecting local representations while minimizing associated utility costs. Our experiments demonstrate that the FeT surpasses the baseline models by up to 46\% in terms of accuracy when scaled to 50 parties. Additionally, in two-party fuzzy VFL settings, FeT also shows improved performance and privacy over cutting-edge VFL models.

☆ Stick-breaking Attention

Shawn Tan, Yikang Shen, Songlin Yang, Aaron Courville, Rameswar Panda

The self-attention mechanism traditionally relies on the softmax operator, necessitating positional embeddings like RoPE, or position biases to account for token order. But current methods using still face length generalisation challenges. We propose an alternative attention mechanism based on the stick-breaking process: For each token before the current, we determine a break point $\beta_{i,j}$, which represents the proportion of the remaining stick to allocate to the current token. We repeat the process until the stick is fully allocated, resulting in a sequence of attention weights. This process naturally incorporates recency bias, which has linguistic motivations for grammar parsing (Shen et. al., 2017). We study the implications of replacing the conventional softmax-based attention mechanism with stick-breaking attention. We then discuss implementation of numerically stable stick-breaking attention and adapt Flash Attention to accommodate this mechanism. When used as a drop-in replacement for current softmax+RoPE attention systems, we find that stick-breaking attention performs competitively with current methods on length generalisation and downstream tasks. Stick-breaking also performs well at length generalisation, allowing a model trained with $2^{11}$ context window to perform well at $2^{14}$ with perplexity improvements.

☆ metasnf: Meta Clustering with Similarity Network Fusion in R

Prashanth S Velayudhan, Xiaoqiao Xu, Prajkta Kallurkar, Ana Patricia Balbon, Maria T Secara, Adam Taback, Denise Sabac, Nicholas Chan, Shihao Ma, Bo Wang, Daniel Felsky, Stephanie H Ameis, Brian Cox, Colin Hawco, Lauren Erdman, Anne L Wheeler

metasnf is an R package that enables users to apply meta clustering, a method for efficiently searching a broad space of cluster solutions by clustering the solutions themselves, to clustering workflows based on similarity network fusion (SNF). SNF is a multi-modal data integration algorithm commonly used for biomedical subtype discovery. The package also contains functions to assist with cluster visualization, characterization, and validation. This package can help researchers identify SNF-derived cluster solutions that are guided by context-specific utility over context-agnostic measures of quality.

comment: 72 pages, 22 figures, submitted to Journal of Statistical Software

☆ Optical Generative Models

Shiqi Chen, Yuhang Li, Hanlong Chen, Aydogan Ozcan

Generative models cover various application areas, including image, video and music synthesis, natural language processing, and molecular design, among many others. As digital generative models become larger, scalable inference in a fast and energy-efficient manner becomes a challenge. Here, we present optical generative models inspired by diffusion models, where a shallow and fast digital encoder first maps random noise into phase patterns that serve as optical generative seeds for a desired data distribution; a jointly-trained free-space-based reconfigurable decoder all-optically processes these generative seeds to create novel images (never seen before) following the target data distribution. Except for the illumination power and the random seed generation through a shallow encoder, these optical generative models do not consume computing power during the synthesis of novel images. We report the optical generation of monochrome and multi-color novel images of handwritten digits, fashion products, butterflies, and human faces, following the data distributions of MNIST, Fashion MNIST, Butterflies-100, and Celeb-A datasets, respectively, achieving an overall performance comparable to digital neural network-based generative models. To experimentally demonstrate optical generative models, we used visible light to generate, in a snapshot, novel images of handwritten digits and fashion products. These optical generative models might pave the way for energy-efficient, scalable and rapid inference tasks, further exploiting the potentials of optics and photonics for artificial intelligence-generated content.

comment: 24 Pages, 9 Figures

☆ POMDP-Driven Cognitive Massive MIMO Radar: Joint Target Detection-Tracking In Unknown Disturbances

Imad Bouhou, Stefano Fortunati, Leila Gharsalli, Alexandre Renaux

The joint detection and tracking of a moving target embedded in an unknown disturbance represents a key feature that motivates the development of the cognitive radar paradigm. Building upon recent advancements in robust target detection with multiple-input multiple-output (MIMO) radars, this work explores the application of a Partially Observable Markov Decision Process (POMDP) framework to enhance the tracking and detection tasks in a statistically unknown environment. In the POMDP setup, the radar system is considered as an intelligent agent that continuously senses the surrounding environment, optimizing its actions to maximize the probability of detection $(P_D)$ and improve the target position and velocity estimation, all this while keeping a constant probability of false alarm $(P_{FA})$. The proposed approach employs an online algorithm that does not require any apriori knowledge of the noise statistics, and it relies on a much more general observation model than the traditional range-azimuth-elevation model employed by conventional tracking algorithms. Simulation results clearly show substantial performance improvement of the POMDP-based algorithm compared to the State-Action-Reward-State-Action (SARSA)-based one that has been recently investigated in the context of massive MIMO (MMIMO) radar systems.

comment: The paper has been submitted to ieee Transactions on radar systems

☆ A Time-Aware Approach to Early Detection of Anorexia: UNSL at eRisk 2024

Horacio Thompson, Marcelo Errecalde

The eRisk laboratory aims to address issues related to early risk detection on the Web. In this year's edition, three tasks were proposed, where Task 2 was about early detection of signs of anorexia. Early risk detection is a problem where precision and speed are two crucial objectives. Our research group solved Task 2 by defining a CPI+DMC approach, addressing both objectives independently, and a time-aware approach, where precision and speed are considered a combined single-objective. We implemented the last approach by explicitly integrating time during the learning process, considering the ERDE{\theta} metric as the training objective. It also allowed us to incorporate temporal metrics to validate and select the optimal models. We achieved outstanding results for the ERDE50 metric and ranking-based metrics, demonstrating consistency in solving ERD problems.

comment: In Conference and Labs of the Evaluation Forum (CLEF 2024), Grenoble, France

☆ Closed-form merging of parameter-efficient modules for Federated Continual Learning

Riccardo Salami, Pietro Buzzega, Matteo Mosconi, Jacopo Bonato, Luigi Sabetta, Simone Calderara

Model merging has emerged as a crucial technique in Deep Learning, enabling the integration of multiple models into a unified system while preserving performance and scalability. In this respect, the compositional properties of low-rank adaptation techniques (e.g., LoRA) have proven beneficial, as simple averaging LoRA modules yields a single model that mostly integrates the capabilities of all individual modules. Building on LoRA, we take a step further by imposing that the merged model matches the responses of all learned modules. Solving this objective in closed form yields an indeterminate system with A and B as unknown variables, indicating the existence of infinitely many closed-form solutions. To address this challenge, we introduce LoRM, an alternating optimization strategy that trains one LoRA matrix at a time. This allows solving for each unknown variable individually, thus finding a unique solution. We apply our proposed methodology to Federated Class-Incremental Learning (FCIL), ensuring alignment of model responses both between clients and across tasks. Our method demonstrates state-of-the-art performance across a range of FCIL scenarios.

☆ Medical Imaging Complexity and its Effects on GAN Performance ACCV

William Cagas, Chan Ko, Blake Hsiao, Shryuk Grandhi, Rishi Bhattacharya, Kevin Zhu, Michael Lam

The proliferation of machine learning models in diverse clinical applications has led to a growing need for high-fidelity, medical image training data. Such data is often scarce due to cost constraints and privacy concerns. Alleviating this burden, medical image synthesis via generative adversarial networks (GANs) emerged as a powerful method for synthetically generating photo-realistic images based on existing sets of real medical images. However, the exact image set size required to efficiently train such a GAN is unclear. In this work, we experimentally establish benchmarks that measure the relationship between a sample dataset size and the fidelity of the generated images, given the dataset's distribution of image complexities. We analyze statistical metrics based on delentropy, an image complexity measure rooted in Shannon's entropy in information theory. For our pipeline, we conduct experiments with two state-of-the-art GANs, StyleGAN 3 and SPADE-GAN, trained on multiple medical imaging datasets with variable sample sizes. Across both GANs, general performance improved with increasing training set size but suffered with increasing complexity.

comment: Accepted to ACCV, Workshop on Generative AI for Synthetic Medical Data

☆ MCUBERT: Memory-Efficient BERT Inference on Commodity Microcontrollers

Zebin Yang, Renze Chen, Taiqiang Wu, Ngai Wong, Yun Liang, Runsheng Wang, Ru Huang, Meng Li

In this paper, we propose MCUBERT to enable language models like BERT on tiny microcontroller units (MCUs) through network and scheduling co-optimization. We observe the embedding table contributes to the major storage bottleneck for tiny BERT models. Hence, at the network level, we propose an MCU-aware two-stage neural architecture search algorithm based on clustered low-rank approximation for embedding compression. To reduce the inference memory requirements, we further propose a novel fine-grained MCU-friendly scheduling strategy. Through careful computation tiling and re-ordering as well as kernel design, we drastically increase the input sequence lengths supported on MCUs without any latency or accuracy penalty. MCUBERT reduces the parameter size of BERT-tiny and BERT-mini by 5.7$\times$ and 3.0$\times$ and the execution memory by 3.5$\times$ and 4.3$\times$, respectively. MCUBERT also achieves 1.5$\times$ latency reduction. For the first time, MCUBERT enables lightweight BERT models on commodity MCUs and processing more than 512 tokens with less than 256KB of memory.

comment: ICCAD 2024

☆ SimRAG: Self-Improving Retrieval-Augmented Generation for Adapting Large Language Models to Specialized Domains

Ran Xu, Hui Liu, Sreyashi Nag, Zhenwei Dai, Yaochen Xie, Xianfeng Tang, Chen Luo, Yang Li, Joyce C. Ho, Carl Yang, Qi He

Retrieval-augmented generation (RAG) enhances the question-answering (QA) abilities of large language models (LLMs) by integrating external knowledge. However, adapting general-purpose RAG systems to specialized fields such as science and medicine poses unique challenges due to distribution shifts and limited access to domain-specific data. To tackle this, we propose SimRAG, a self-training approach that equips the LLM with joint capabilities of question answering and question generation for domain adaptation. Our method first fine-tunes the LLM on instruction-following, question-answering, and search-related data. Then, it prompts the same LLM to generate diverse domain-relevant questions from unlabeled corpora, with an additional filtering strategy to retain high-quality synthetic examples. By leveraging these synthetic examples, the LLM can improve their performance on domain-specific RAG tasks. Experiments on 11 datasets, spanning two backbone sizes and three domains, demonstrate that SimRAG outperforms baselines by 1.2\%--8.6\%.

comment: Work in Progress

☆ Generalized Resubstitution for Regression Error Estimation

Diego Marcondes, Ulisses Braga-Neto

We propose generalized resubstitution error estimators for regression, a broad family of estimators, each corresponding to a choice of empirical probability measures and loss function. The usual sum of squares criterion is a special case corresponding to the standard empirical probability measure and the quadratic loss. Other choices of empirical probability measure lead to more general estimators with superior bias and variance properties. We prove that these error estimators are consistent under broad assumptions. In addition, procedures for choosing the empirical measure based on the method of moments and maximum pseudo-likelihood are proposed and investigated. Detailed experimental results using polynomial regression demonstrate empirically the superior finite-sample bias and variance properties of the proposed estimators. The R code for the experiments is provided.

☆ Theoretically Grounded Pruning of Large Ground Sets for Constrained, Discrete Optimization

Ankur Nath, Alan Kuhnle

Modern instances of combinatorial optimization problems often exhibit billion-scale ground sets, which have many uninformative or redundant elements. In this work, we develop light-weight pruning algorithms to quickly discard elements that are unlikely to be part of an optimal solution. Under mild assumptions on the instance, we prove theoretical guarantees on the fraction of the optimal value retained and the size of the resulting pruned ground set. Through extensive experiments on real-world datasets for various applications, we demonstrate that our algorithm, QuickPrune, efficiently prunes over 90% of the ground set and outperforms state-of-the-art classical and machine learning heuristics for pruning.

☆ Optimizing Travel Itineraries with AI Algorithms in a Microservices Architecture: Balancing Cost, Time, Preferences, and Sustainability

Biman Barua, M. Shamim Kaiser

The objective of this research is how an implementation of AI algorithms in the microservices architecture enhances travel itineraries by cost, time, user preferences, and environmental sustainability. It uses machine learning models for both cost forecasting and personalization, genetic algorithm for optimization of the itinerary, and heuristics for sustainability checking. Primary evaluated parameters consist of latency, ability to satisfy user preferences, cost and environmental concern. The experimental results demonstrate an average of 4.5 seconds of response time on 1000 concurrent users and 92% of user preferences accuracy. The cost efficiency is proved, with 95% of provided trips being within the limits of the budget declared by the user. The system also implements some measures to alleviate negative externalities related to travel and 60% of offered travel plans had green options incorporated, resulting in the average 15% lower carbon emissions than the traditional travel plans offered. The genetic algorithm with time complexity O(g.p.f) provides the optimal solution in 100 generations. Every iteration improves the quality of the solution by 5%, thus enabling its effective use in optimization problems where time is measured in seconds. Finally, the system is designed to be fault-tolerant with functional 99.9% availability which allows the provision of services even when requirements are exceeded. Travel optimization platform is turned dynamic and efficient by this microservices based architecture which provides enhanced scaling, allows asynchronous communication and real time changes. Because of the incorporation of Ai, cost control and eco-friendliness approaches, the system addresses the different user needs in the present days travel business.

comment: 18 pages, 6 figures

☆ Spiking Graph Neural Network on Riemannian Manifolds NeurIPS 2024

Li Sun, Zhenhao Huang, Qiqi Wan, Hao Peng, Philip S. Yu

Graph neural networks (GNNs) have become the dominant solution for learning on graphs, the typical non-Euclidean structures. Conventional GNNs, constructed with the Artificial Neuron Network (ANN), have achieved impressive performance at the cost of high computation and energy consumption. In parallel, spiking GNNs with brain-like spiking neurons are drawing increasing research attention owing to the energy efficiency. So far, existing spiking GNNs consider graphs in Euclidean space, ignoring the structural geometry, and suffer from the high latency issue due to Back-Propagation-Through-Time (BPTT) with the surrogate gradient. In light of the aforementioned issues, we are devoted to exploring spiking GNN on Riemannian manifolds, and present a Manifold-valued Spiking GNN (MSG). In particular, we design a new spiking neuron on geodesically complete manifolds with the diffeomorphism, so that BPTT regarding the spikes is replaced by the proposed differentiation via manifold. Theoretically, we show that MSG approximates a solver of the manifold ordinary differential equation. Extensive experiments on common graphs show the proposed MSG achieves superior performance to previous spiking GNNs and energy efficiency to conventional GNNs.

comment: Accepted by NeurIPS 2024, 30 pages

☆ Semi-Implicit Functional Gradient Flow

Shiyue Zhang, Ziheng Cheng, Cheng Zhang

Particle-based variational inference methods (ParVIs) use non-parametric variational families represented by particles to approximate the target distribution according to the kernelized Wasserstein gradient flow for the Kullback-Leibler (KL) divergence. Recent works introduce functional gradient flows to substitute the kernel for better flexibility. However, the deterministic updating mechanism may suffer from limited exploration and require expensive repetitive runs for new samples. In this paper, we propose Semi-Implicit Functional Gradient flow (SIFG), a functional gradient ParVI method that uses perturbed particles as the approximation family. The corresponding functional gradient flow, which can be estimated via denoising score matching, exhibits strong theoretical convergence guarantee. We also present an adaptive version of our method to automatically choose the suitable noise magnitude. Extensive experiments demonstrate the effectiveness and efficiency of the proposed framework on both simulated and real data problems.

comment: 31 pages, 12 figures

☆ Retrieving snow depth distribution by downscaling ERA5 Reanalysis with ICESat-2 laser altimetry

Zhihao Liu, Simon Filhol, Désirée Treichler

Estimating the variability of seasonal snow cover, in particular snow depth in remote areas, poses significant challenges due to limited spatial and temporal data availability. This study uses snow depth measurements from the ICESat-2 satellite laser altimeter, which are sparse in both space and time, and incorporates them with climate reanalysis data into a downscaling-calibration scheme to produce monthly gridded snow depth maps at microscale (10 m). Snow surface elevation measurements from ICESat-2 along profiles are compared to a digital elevation model to determine snow depth at each point. To efficiently turn sparse measurements into snow depth maps, a regression model is fitted to establish a relationship between the retrieved snow depth and the corresponding ERA5 Land snow depth. This relationship, referred to as subgrid variability, is then applied to downscale the monthly ERA5 Land snow depth data. The method can provide timeseries of monthly snow depth maps for the entire ERA5 time range (since 1950). The validation of downscaled snow depth data was performed at an intermediate scale (100 m x 500 m) using datasets from airborne laser scanning (ALS) in the Hardangervidda region of southern Norway. Results show that snow depth prediction achieved R2 values ranging from 0.74 to 0.88 (post-calibration). The method relies on globally available data and is applicable to other snow regions above the treeline. Though requiring area-specific calibration, our approach has the potential to provide snow depth maps in areas where no such data exist and can be used to extrapolate existing snow surveys in time and over larger areas. With this, it can offer valuable input data for hydrological, ecological or permafrost modeling tasks.

☆ Multi-Continental Healthcare Modelling Using Blockchain-Enabled Federated Learning

Rui Sun, Zhipeng Wang, Hengrui Zhang, Ming Jiang, Yizhe Wen, Jiqun Zhang, Jiahao Sun, Shuoying Zhang, Erwu Liu, Kezhi Li

One of the biggest challenges of building artificial intelligence (AI) model in healthcare area is the data sharing. Since healthcare data is private, sensitive, and heterogeneous, collecting sufficient data for modelling is exhausted, costly, and sometimes impossible. In this paper, we propose a framework for global healthcare modelling using datasets from multi-continents (Europe, North America and Asia) while without sharing the local datasets, and choose glucose management as a study model to verify its effectiveness. Technically, blockchain-enabled federated learning is implemented with adaption to make it meet with the privacy and safety requirements of healthcare data, meanwhile rewards honest participation and penalize malicious activities using its on-chain incentive mechanism. Experimental results show that the proposed framework is effective, efficient, and privacy preserved. Its prediction accuracy is much better than the models trained from limited personal data and is similar to, and even slightly better than, the results from a centralized dataset. This work paves the way for international collaborations on healthcare projects, where additional data is crucial for reducing bias and providing benefits to humanity.

comment: Accepted by IEEE Global Blockchain Conference

☆ Addressing Asynchronicity in Clinical Multimodal Fusion via Individualized Chest X-ray Generation NeurIPS-24

Wenfang Yao, Chen Liu, Kejing Yin, William K. Cheung, Jing Qin

Integrating multi-modal clinical data, such as electronic health records (EHR) and chest X-ray images (CXR), is particularly beneficial for clinical prediction tasks. However, in a temporal setting, multi-modal data are often inherently asynchronous. EHR can be continuously collected but CXR is generally taken with a much longer interval due to its high cost and radiation dose. When clinical prediction is needed, the last available CXR image might have been outdated, leading to suboptimal predictions. To address this challenge, we propose DDL-CXR, a method that dynamically generates an up-to-date latent representation of the individualized CXR images. Our approach leverages latent diffusion models for patient-specific generation strategically conditioned on a previous CXR image and EHR time series, providing information regarding anatomical structures and disease progressions, respectively. In this way, the interaction across modalities could be better captured by the latent CXR generation process, ultimately improving the prediction performance. Experiments using MIMIC datasets show that the proposed model could effectively address asynchronicity in multimodal fusion and consistently outperform existing methods.

comment: Accepted by NeurIPS-24

☆ regAL: Python Package for Active Learning of Regression Problems

Elizaveta Surzhikova, Jonny Proppe

Increasingly more research areas rely on machine learning methods to accelerate discovery while saving resources. Machine learning models, however, usually require large datasets of experimental or computational results, which in certain fields, such as (bio)chemistry, materials science, or medicine, are rarely given and often prohibitively expensive to obtain. To bypass that obstacle, active learning methods are employed to develop machine learning models with a desired performance while requiring the least possible number of computational or experimental results from the domain of application. For this purpose, the model's knowledge about certain regions of the application domain is estimated to guide the choice of the model's training set. Although active learning is widely studied for classification problems (discrete outcomes), comparatively few works handle this method for regression problems (continuous outcomes). In this work, we present our Python package regAL, which allows users to evaluate different active learning strategies for regression problems. With a minimal input of just the dataset in question, but many additional customization and insight options, this package is intended for anyone who aims to perform and understand active learning in their problem-specific scope.

☆ Deep learning for model correction of dynamical systems with data scarcity

Caroline Tatsuoka, Dongbin Xiu

We present a deep learning framework for correcting existing dynamical system models utilizing only a scarce high-fidelity data set. In many practical situations, one has a low-fidelity model that can capture the dynamics reasonably well but lacks high resolution, due to the inherent limitation of the model and the complexity of the underlying physics. When high resolution data become available, it is natural to seek model correction to improve the resolution of the model predictions. We focus on the case when the amount of high-fidelity data is so small that most of the existing data driven modeling methods cannot be applied. In this paper, we address these challenges with a model-correction method which only requires a scarce high-fidelity data set. Our method first seeks a deep neural network (DNN) model to approximate the existing low-fidelity model. By using the scarce high-fidelity data, the method then corrects the DNN model via transfer learning (TL). After TL, an improved DNN model with high prediction accuracy to the underlying dynamics is obtained. One distinct feature of the propose method is that it does not assume a specific form of the model correction terms. Instead, it offers an inherent correction to the low-fidelity model via TL. A set of numerical examples are presented to demonstrate the effectiveness of the proposed method.

☆ Reinforcement Learning under Latent Dynamics: Toward Statistical and Algorithmic Modularity

Philip Amortila, Dylan J. Foster, Nan Jiang, Akshay Krishnamurthy, Zakaria Mhammedi

Real-world applications of reinforcement learning often involve environments where agents operate on complex, high-dimensional observations, but the underlying (''latent'') dynamics are comparatively simple. However, outside of restrictive settings such as small latent spaces, the fundamental statistical requirements and algorithmic principles for reinforcement learning under latent dynamics are poorly understood. This paper addresses the question of reinforcement learning under $\textit{general}$ latent dynamics from a statistical and algorithmic perspective. On the statistical side, our main negative result shows that most well-studied settings for reinforcement learning with function approximation become intractable when composed with rich observations; we complement this with a positive result, identifying latent pushforward coverability as a general condition that enables statistical tractability. Algorithmically, we develop provably efficient observable-to-latent reductions -- that is, reductions that transform an arbitrary algorithm for the latent MDP into an algorithm that can operate on rich observations -- in two settings: one where the agent has access to hindsight observations of the latent dynamics [LADZ23], and one where the agent can estimate self-predictive latent models [SAGHCB20]. Together, our results serve as a first step toward a unified statistical and algorithmic theory for reinforcement learning under latent dynamics.

☆ Scalable Offline Reinforcement Learning for Mean Field Games AAMAS

Axel Brunnbauer, Julian Lemmel, Zahra Babaiee, Sophie Neubauer, Radu Grosu

Reinforcement learning algorithms for mean-field games offer a scalable framework for optimizing policies in large populations of interacting agents. Existing methods often depend on online interactions or access to system dynamics, limiting their practicality in real-world scenarios where such interactions are infeasible or difficult to model. In this paper, we present Offline Munchausen Mirror Descent (Off-MMD), a novel mean-field RL algorithm that approximates equilibrium policies in mean-field games using purely offline data. By leveraging iterative mirror descent and importance sampling techniques, Off-MMD estimates the mean-field distribution from static datasets without relying on simulation or environment dynamics. Additionally, we incorporate techniques from offline reinforcement learning to address common issues like Q-value overestimation, ensuring robust policy learning even with limited data coverage. Our algorithm scales to complex environments and demonstrates strong performance on benchmark tasks like crowd exploration or navigation, highlighting its applicability to real-world multi-agent systems where online experimentation is infeasible. We empirically demonstrate the robustness of Off-MMD to low-quality datasets and conduct experiments to investigate its sensitivity to hyperparameter choices.

comment: Submitted to AAMAS

☆ Identifiable Representation and Model Learning for Latent Dynamic Systems

Congxi Zhang, Yongchun Xie

Learning identifiable representations and models from low-level observations is useful for an intelligent spacecraft to reliability finish downstream tasks. For temporal observations, to ensure that the data generating process is provably inverted, most existing works either assume the noise variables in the dynamic mechanisms are (conditionally) independent, or require interventions which can directly affect each latent variable. However, in practice, the relationship between the exogenous inputs/interventions and the latent variables may follow some complex deterministic mechanisms. In this work, we study the problem of identifiable representation and model learning for latent dynamic systems. The key idea is that we use an inductive bias inspired by controllable canonical forms, which is invariant, sparse, and input dependent by definition. We prove that, for linear or affine nonlinear latent dynamic systems, it is possible to identify the representations up to scaling and determine the models up to some simple transformations. The results have potential to provide some theoretical guarantees for developing more trustworthy decision-making and control methods for intelligent spacecrafts.

☆ AdaRankGrad: Adaptive Gradient-Rank and Moments for Memory-Efficient LLMs Training and Fine-Tuning

Yehonathan Refael, Jonathan Svirsky, Boris Shustin, Wasim Huleihel, Ofir Lindenbaum

Training and fine-tuning large language models (LLMs) come with challenges related to memory and computational requirements due to the increasing size of the model weights and the optimizer states. Various techniques have been developed to tackle these challenges, such as low-rank adaptation (LoRA), which involves introducing a parallel trainable low-rank matrix to the fixed pre-trained weights at each layer. However, these methods often fall short compared to the full-rank weight training approach, as they restrict the parameter search to a low-rank subspace. This limitation can disrupt training dynamics and require a full-rank warm start to mitigate the impact. In this paper, we introduce a new method inspired by a phenomenon we formally prove: as training progresses, the rank of the estimated layer gradients gradually decreases, and asymptotically approaches rank one. Leveraging this, our approach involves adaptively reducing the rank of the gradients during Adam optimization steps, using an efficient online-updating low-rank projections rule. We further present a randomized SVD scheme for efficiently finding the projection matrix. Our technique enables full-parameter fine-tuning with adaptive low-rank gradient updates, significantly reducing overall memory requirements during training compared to state-of-the-art methods while improving model performance in both pretraining and fine-tuning. Finally, we provide a convergence analysis of our method and demonstrate its merits for training and fine-tuning language and biological foundation models.

☆ Relaxed Equivariance via Multitask Learning

Ahmed A. Elhag, T. Konstantin Rusch, Francesco Di Giovanni, Michael Bronstein

Incorporating equivariance as an inductive bias into deep learning architectures to take advantage of the data symmetry has been successful in multiple applications, such as chemistry and dynamical systems. In particular, roto-translations are crucial for effectively modeling geometric graphs and molecules, where understanding the 3D structures enhances generalization. However, equivariant models often pose challenges due to their high computational complexity. In this paper, we introduce REMUL, a training procedure for approximating equivariance with multitask learning. We show that unconstrained models (which do not build equivariance into the architecture) can learn approximate symmetries by minimizing an additional simple equivariance loss. By formulating equivariance as a new learning objective, we can control the level of approximate equivariance in the model. Our method achieves competitive performance compared to equivariant baselines while being $10 \times$ faster at inference and $2.5 \times$ at training.

☆ Population stratification for prediction of mortality in post-AKI patients

Flavio S. Correa da Silva, Simon Sawhney

Acute kidney injury (AKI) is a serious clinical condition that affects up to 20% of hospitalised patients. AKI is associated with short term unplanned hospital readmission and post-discharge mortality risk. Patient risk and healthcare expenditures can be minimised by followup planning grounded on predictive models and machine learning. Since AKI is multi-factorial, predictive models specialised in different categories of patients can increase accuracy of predictions. In the present article we present some results following this approach.

☆ CASCRNet: An Atrous Spatial Pyramid Pooling and Shared Channel Residual based Network for Capsule Endoscopy

K V Srinanda, M Manvith Prabhu, Shyam Lal

This manuscript summarizes work on the Capsule Vision Challenge 2024 by MISAHUB. To address the multi-class disease classification task, which is challenging due to the complexity and imbalance in the Capsule Vision challenge dataset, this paper proposes CASCRNet (Capsule endoscopy-Aspp-SCR-Network), a parameter-efficient and novel model that uses Shared Channel Residual (SCR) blocks and Atrous Spatial Pyramid Pooling (ASPP) blocks. Further, the performance of the proposed model is compared with other well-known approaches. The experimental results yield that proposed model provides better disease classification results. The proposed model was successful in classifying diseases with an F1 Score of 78.5% and a Mean AUC of 98.3%, which is promising given its compact architecture.

comment: 8 pages, 4 figures

☆ The Probabilistic Tsetlin Machine: A Novel Approach to Uncertainty Quantification

K. Darshana Abeyrathna, Sara El Mekkaoui, Andreas Hafver, Christian Agrell

Tsetlin Machines (TMs) have emerged as a compelling alternative to conventional deep learning methods, offering notable advantages such as smaller memory footprint, faster inference, fault-tolerant properties, and interpretability. Although various adaptations of TMs have expanded their applicability across diverse domains, a fundamental gap remains in understanding how TMs quantify uncertainty in their predictions. In response, this paper introduces the Probabilistic Tsetlin Machine (PTM) framework, aimed at providing a robust, reliable, and interpretable approach for uncertainty quantification. Unlike the original TM, the PTM learns the probability of staying on each state of each Tsetlin Automaton (TA) across all clauses. These probabilities are updated using the feedback tables that are part of the TM framework: Type I and Type II feedback. During inference, TAs decide their actions by sampling states based on learned probability distributions, akin to Bayesian neural networks when generating weight values. In our experimental analysis, we first illustrate the spread of the probabilities across TA states for the noisy-XOR dataset. Then we evaluate the PTM alongside benchmark models using both simulated and real-world datasets. The experiments on the simulated dataset reveal the PTM's effectiveness in uncertainty quantification, particularly in delineating decision boundaries and identifying regions of high uncertainty. Moreover, when applied to multiclass classification tasks using the Iris dataset, the PTM demonstrates competitive performance in terms of predictive entropy and expected calibration error, showcasing its potential as a reliable tool for uncertainty estimation. Our findings underscore the importance of selecting appropriate models for accurate uncertainty quantification in predictive tasks, with the PTM offering a particularly interpretable and effective solution.

comment: 12 pages, 5 figures, 6 tables, accepted and presented at ICAAI 2024, London

☆ Is the GPU Half-Empty or Half-Full? Practical Scheduling Techniques for LLMs

Ferdi Kossmann, Bruce Fontaine, Daya Khudia, Michael Cafarella, Samuel Madden

Serving systems for Large Language Models (LLMs) improve throughput by processing several requests concurrently. However, multiplexing hardware resources between concurrent requests involves non-trivial scheduling decisions. Practical serving systems typically implement these decisions at two levels: First, a load balancer routes requests to different servers which each hold a replica of the LLM. Then, on each server, an engine-level scheduler decides when to run a request, or when to queue or preempt it. Improved scheduling policies may benefit a wide range of LLM deployments and can often be implemented as "drop-in replacements" to a system's current policy. In this work, we survey scheduling techniques from the literature and from practical serving systems. We find that schedulers from the literature often achieve good performance but introduce significant complexity. In contrast, schedulers in practical deployments often leave easy performance gains on the table but are easy to implement, deploy and configure. This finding motivates us to introduce two new scheduling techniques, which are both easy to implement, and outperform current techniques on production workload traces.

comment: 12 pages, 11 figures

☆ Optimal Streaming Algorithms for Multi-Armed Bandits

Tianyuan Jin, Keke Huang, Jing Tang, Xiaokui Xiao

This paper studies two variants of the best arm identification (BAI) problem under the streaming model, where we have a stream of $n$ arms with reward distributions supported on $[0,1]$ with unknown means. The arms in the stream are arriving one by one, and the algorithm cannot access an arm unless it is stored in a limited size memory. We first study the streaming \eps-$top$-$k$ arms identification problem, which asks for $k$ arms whose reward means are lower than that of the $k$-th best arm by at most $\eps$ with probability at least $1-\delta$. For general $\eps \in (0,1)$, the existing solution for this problem assumes $k = 1$ and achieves the optimal sample complexity $O(\frac{n}{\eps^2} \log \frac{1}{\delta})$ using $O(\log^*(n))$ ($\log^*(n)$ equals the number of times that we need to apply the logarithm function on $n$ before the results is no more than 1.) memory and a single pass of the stream. We propose an algorithm that works for any $k$ and achieves the optimal sample complexity $O(\frac{n}{\eps^2} \log\frac{k}{\delta})$ using a single-arm memory and a single pass of the stream. Second, we study the streaming BAI problem, where the objective is to identify the arm with the maximum reward mean with at least $1-\delta$ probability, using a single-arm memory and as few passes of the input stream as possible. We present a single-arm-memory algorithm that achieves a near instance-dependent optimal sample complexity within $O(\log \Delta_2^{-1})$ passes, where $\Delta_2$ is the gap between the mean of the best arm and that of the second best arm.

comment: 24pages

☆ Non-intrusive Speech Quality Assessment with Diffusion Models Trained on Clean Speech

Danilo de Oliveira, Julius Richter, Jean-Marie Lemercier, Simon Welker, Timo Gerkmann

Diffusion models have found great success in generating high quality, natural samples of speech, but their potential for density estimation for speech has so far remained largely unexplored. In this work, we leverage an unconditional diffusion model trained only on clean speech for the assessment of speech quality. We show that the quality of a speech utterance can be assessed by estimating the likelihood of a corresponding sample in the terminating Gaussian distribution, obtained via a deterministic noising process. The resulting method is purely unsupervised, trained only on clean speech, and therefore does not rely on annotations. Our diffusion-based approach leverages clean speech priors to assess quality based on how the input relates to the learned distribution of clean data. Our proposed log-likelihoods show promising results, correlating well with intrusive speech quality metrics such as POLQA and SI-SDR.

☆ Att2CPC: Attention-Guided Lossy Attribute Compression of Point Clouds

Kai Liu, Kang You, Pan Gao, Manoranjan Paul

With the great progress of 3D sensing and acquisition technology, the volume of point cloud data has grown dramatically, which urges the development of efficient point cloud compression methods. In this paper, we focus on the task of learned lossy point cloud attribute compression (PCAC). We propose an efficient attention-based method for lossy compression of point cloud attributes leveraging on an autoencoder architecture. Specifically, at the encoding side, we conduct multiple downsampling to best exploit the local attribute patterns, in which effective External Cross Attention (ECA) is devised to hierarchically aggregate features by intergrating attributes and geometry contexts. At the decoding side, the attributes of the point cloud are progressively reconstructed based on the multi-scale representation and the zero-padding upsampling tactic. To the best of our knowledge, this is the first approach to introduce attention mechanism to point-based lossy PCAC task. We verify the compression efficiency of our model on various sequences, including human body frames, sparse objects, and large-scale point cloud scenes. Experiments show that our method achieves an average improvement of 1.15 dB and 2.13 dB in BD-PSNR of Y channel and YUV channel, respectively, when comparing with the state-of-the-art point-based method Deep-PCAC. Codes of this paper are available at https://github.com/I2-Multimedia-Lab/Att2CPC.

☆ Learning Lossless Compression for High Bit-Depth Volumetric Medical Image

Kai Wang, Yuanchao Bai, Daxin Li, Deming Zhai, Junjun Jiang, Xianming Liu

Recent advances in learning-based methods have markedly enhanced the capabilities of image compression. However, these methods struggle with high bit-depth volumetric medical images, facing issues such as degraded performance, increased memory demand, and reduced processing speed. To address these challenges, this paper presents the Bit-Division based Lossless Volumetric Image Compression (BD-LVIC) framework, which is tailored for high bit-depth medical volume compression. The BD-LVIC framework skillfully divides the high bit-depth volume into two lower bit-depth segments: the Most Significant Bit-Volume (MSBV) and the Least Significant Bit-Volume (LSBV). The MSBV concentrates on the most significant bits of the volumetric medical image, capturing vital structural details in a compact manner. This reduction in complexity greatly improves compression efficiency using traditional codecs. Conversely, the LSBV deals with the least significant bits, which encapsulate intricate texture details. To compress this detailed information effectively, we introduce an effective learning-based compression model equipped with a Transformer-Based Feature Alignment Module, which exploits both intra-slice and inter-slice redundancies to accurately align features. Subsequently, a Parallel Autoregressive Coding Module merges these features to precisely estimate the probability distribution of the least significant bit-planes. Our extensive testing demonstrates that the BD-LVIC framework not only sets new performance benchmarks across various datasets but also maintains a competitive coding speed, highlighting its significant potential and practical utility in the realm of volumetric medical image compression.

comment: 13 pages

☆ A Comprehensive Analysis on the Learning Curve in Kernel Ridge Regression

Tin Sum Cheng, Aurelien Lucchi, Anastasis Kratsios, David Belius

This paper conducts a comprehensive study of the learning curves of kernel ridge regression (KRR) under minimal assumptions. Our contributions are three-fold: 1) we analyze the role of key properties of the kernel, such as its spectral eigen-decay, the characteristics of the eigenfunctions, and the smoothness of the kernel; 2) we demonstrate the validity of the Gaussian Equivalent Property (GEP), which states that the generalization performance of KRR remains the same when the whitened features are replaced by standard Gaussian vectors, thereby shedding light on the success of previous analyzes under the Gaussian Design Assumption; 3) we derive novel bounds that improve over existing bounds across a broad range of setting such as (in)dependent feature vectors and various combinations of eigen-decay rates in the over/underparameterized regimes.

☆ Enhancing Federated Learning Convergence with Dynamic Data Queue and Data Entropy-driven Participant Selection

Charuka Herath, Xiaolan Liu, Sangarapillai Lambotharan, Yogachandran Rahulamathavan

Federated Learning (FL) is a decentralized approach for collaborative model training on edge devices. This distributed method of model training offers advantages in privacy, security, regulatory compliance, and cost-efficiency. Our emphasis in this research lies in addressing statistical complexity in FL, especially when the data stored locally across devices is not identically and independently distributed (non-IID). We have observed an accuracy reduction of up to approximately 10\% to 30\%, particularly in skewed scenarios where each edge device trains with only 1 class of data. This reduction is attributed to weight divergence, quantified using the Euclidean distance between device-level class distributions and the population distribution, resulting in a bias term (\(\delta_k\)). As a solution, we present a method to improve convergence in FL by creating a global subset of data on the server and dynamically distributing it across devices using a Dynamic Data queue-driven Federated Learning (DDFL). Next, we leverage Data Entropy metrics to observe the process during each training round and enable reasonable device selection for aggregation. Furthermore, we provide a convergence analysis of our proposed DDFL to justify their viability in practical FL scenarios, aiming for better device selection, a non-sub-optimal global model, and faster convergence. We observe that our approach results in a substantial accuracy boost of approximately 5\% for the MNIST dataset, around 18\% for CIFAR-10, and 20\% for CIFAR-100 with a 10\% global subset of data, outperforming the state-of-the-art (SOTA) aggregation algorithms.

comment: The Journal is submitted to IEEE Transactions in the Internet of Things

☆ Large Language Models Engineer Too Many Simple Features For Tabular Data

Jaris Küken, Lennart Purucker, Frank Hutter

Tabular machine learning problems often require time-consuming and labor-intensive feature engineering. Recent efforts have focused on using large language models (LLMs) to capitalize on their potential domain knowledge. At the same time, researchers have observed ethically concerning negative biases in other LLM-related use cases, such as text generation. These developments motivated us to investigate whether LLMs exhibit a bias that negatively impacts the performance of feature engineering. While not ethically concerning, such a bias could hinder practitioners from fully utilizing LLMs for automated data science. Therefore, we propose a method to detect potential biases by detecting anomalies in the frequency of operators (e.g., adding two features) suggested by LLMs when engineering new features. Our experiments evaluate the bias of four LLMs, two big frontier and two small open-source models, across 27 tabular datasets. Our results indicate that LLMs are biased toward simple operators, such as addition, and can fail to utilize more complex operators, such as grouping followed by aggregations. Furthermore, the bias can negatively impact the predictive performance when using LLM-generated features. Our results call for mitigating bias when using LLMs for feature engineering.

comment: Preprint

☆ Scaling Robot Policy Learning via Zero-Shot Labeling with Foundation Models

Nils Blank, Moritz Reuss, Marcel Rühle, Ömer Erdinç Yağmurlu, Fabian Wenzel, Oier Mees, Rudolf Lioutikov

A central challenge towards developing robots that can relate human language to their perception and actions is the scarcity of natural language annotations in diverse robot datasets. Moreover, robot policies that follow natural language instructions are typically trained on either templated language or expensive human-labeled instructions, hindering their scalability. To this end, we introduce NILS: Natural language Instruction Labeling for Scalability. NILS automatically labels uncurated, long-horizon robot data at scale in a zero-shot manner without any human intervention. NILS combines pretrained vision-language foundation models in order to detect objects in a scene, detect object-centric changes, segment tasks from large datasets of unlabelled interaction data and ultimately label behavior datasets. Evaluations on BridgeV2, Fractal, and a kitchen play dataset show that NILS can autonomously annotate diverse robot demonstrations of unlabeled and unstructured datasets while alleviating several shortcomings of crowdsourced human annotations, such as low data quality and diversity. We use NILS to label over 115k trajectories obtained from over 430 hours of robot data. We open-source our auto-labeling code and generated annotations on our website: http://robottasklabeling.github.io.

comment: Project Website at https://robottasklabeling.github.io/

☆ Locating Information in Large Language Models via Random Matrix Theory

Max Staats, Matthias Thamm, Bernd Rosenow

As large language models (LLMs) become central to AI applications, gaining a deeper understanding of their inner workings is increasingly important. In this work, we analyze the weight matrices of pretrained transformer models -- specifically BERT and Llama -- using random matrix theory (RMT) as a zero-information hypothesis. While randomly initialized weights perfectly agree with RMT predictions, deviations emerge after training, allowing us to locate learned structures within the models. We identify layer-type specific behaviors that are consistent across all blocks and architectures considered. By pinpointing regions that deviate from RMT predictions, we highlight areas of feature learning and confirm this through comparisons with the activation covariance matrices of the corresponding layers. Our method provides a diagnostic tool for identifying relevant regions in transformer weights using only the trained matrices. Additionally, we address the ongoing debate regarding the significance of small singular values in the context of fine-tuning and alignment in LLMs. Our findings reveal that, after fine-tuning, small singular values play a crucial role in the models' capabilities, suggesting that removing them in an already aligned transformer can be detrimental, as it may compromise model alignment.

comment: 17 pages, 14 figures

☆ Faster Language Models with Better Multi-Token Prediction Using Tensor Decomposition

Artem Basharin, Andrei Chertkov, Ivan Oseledets

We propose a new model for multi-token prediction in transformers, aiming to enhance sampling efficiency without compromising accuracy. Motivated by recent work that predicts the probabilities of subsequent tokens using multiple heads, we connect this approach to rank-$1$ canonical tensor decomposition. By generalizing it to a rank-$r$ canonical probability decomposition, we develop an improved model that predicts multiple tokens simultaneously. This model can also be interpreted as a mixture of experts, allowing us to leverage successful techniques from that domain for efficient and robust training. Importantly, the overall overhead for training and sampling remains low. Our method demonstrates significant improvements in inference speed for both text and code generation tasks, proving particularly beneficial within the self-speculative decoding paradigm. It maintains its effectiveness across various model sizes and training epochs, highlighting its robustness and scalability.

☆ Beyond Backpropagation: Optimization with Multi-Tangent Forward Gradients

Katharina Flügel, Daniel Coquelin, Marie Weiel, Achim Streit, Markus Götz

The gradients used to train neural networks are typically computed using backpropagation. While an efficient way to obtain exact gradients, backpropagation is computationally expensive, hinders parallelization, and is biologically implausible. Forward gradients are an approach to approximate the gradients from directional derivatives along random tangents computed by forward-mode automatic differentiation. So far, research has focused on using a single tangent per step. This paper provides an in-depth analysis of multi-tangent forward gradients and introduces an improved approach to combining the forward gradients from multiple tangents based on orthogonal projections. We demonstrate that increasing the number of tangents improves both approximation quality and optimization performance across various tasks.

☆ Anomaly Resilient Temporal QoS Prediction using Hypergraph Convoluted Transformer Network

Suraj Kumar, Soumi Chattopadhyay, Chandranath Adak

Quality-of-Service (QoS) prediction is a critical task in the service lifecycle, enabling precise and adaptive service recommendations by anticipating performance variations over time in response to evolving network uncertainties and user preferences. However, contemporary QoS prediction methods frequently encounter data sparsity and cold-start issues, which hinder accurate QoS predictions and limit the ability to capture diverse user preferences. Additionally, these methods often assume QoS data reliability, neglecting potential credibility issues such as outliers and the presence of greysheep users and services with atypical invocation patterns. Furthermore, traditional approaches fail to leverage diverse features, including domain-specific knowledge and complex higher-order patterns, essential for accurate QoS predictions. In this paper, we introduce a real-time, trust-aware framework for temporal QoS prediction to address the aforementioned challenges, featuring an end-to-end deep architecture called the Hypergraph Convoluted Transformer Network (HCTN). HCTN combines a hypergraph structure with graph convolution over hyper-edges to effectively address high-sparsity issues by capturing complex, high-order correlations. Complementing this, the transformer network utilizes multi-head attention along with parallel 1D convolutional layers and fully connected dense blocks to capture both fine-grained and coarse-grained dynamic patterns. Additionally, our approach includes a sparsity-resilient solution for detecting greysheep users and services, incorporating their unique characteristics to improve prediction accuracy. Trained with a robust loss function resistant to outliers, HCTN demonstrated state-of-the-art performance on the large-scale WSDREAM-2 datasets for response time and throughput.

comment: 16 pages, 12 figures

☆ Topology meets Machine Learning: An Introduction using the Euler Characteristic Transform

Bastian Rieck

This overview article makes the case for how topological concepts can enrich research in machine learning. Using the Euler Characteristic Transform (ECT), a geometrical-topological invariant, as a running example, I present different use cases that result in more efficient models for analyzing point clouds, graphs, and meshes. Moreover, I outline a vision for how topological concepts could be used in the future, comprising (1) the learning of functions on topological spaces, (2) the building of hybrid models that imbue neural networks with knowledge about the topological information in data, and (3) the analysis of qualitative properties of neural networks. With current research already addressing some of these aspects, this article thus serves as an introduction and invitation to this nascent area of research.

☆ Escaping the Forest: Sparse Interpretable Neural Networks for Tabular Data

Salvatore Raieli, Abdulrahman Altahhan, Nathalie Jeanray, Stéphane Gerart, Sebastien Vachenc

Tabular datasets are widely used in scientific disciplines such as biology. While these disciplines have already adopted AI methods to enhance their findings and analysis, they mainly use tree-based methods due to their interpretability. At the same time, artificial neural networks have been shown to offer superior flexibility and depth for rich and complex non-tabular problems, but they are falling behind tree-based models for tabular data in terms of performance and interpretability. Although sparsity has been shown to improve the interpretability and performance of ANN models for complex non-tabular datasets, enforcing sparsity structurally and formatively for tabular data before training the model, remains an open question. To address this question, we establish a method that infuses sparsity in neural networks by utilising attention mechanisms to capture the features' importance in tabular datasets. We show that our models, Sparse TABular NET or sTAB-Net with attention mechanisms, are more effective than tree-based models, reaching the state-of-the-art on biological datasets. They further permit the extraction of insights from these datasets and achieve better performance than post-hoc methods like SHAP.

☆ VISAGE: Video Synthesis using Action Graphs for Surgery MICCAI 2024

Yousef Yeganeh, Rachmadio Lazuardi, Amir Shamseddin, Emine Dari, Yash Thirani, Nassir Navab Azade Farshad

Surgical data science (SDS) is a field that analyzes patient data before, during, and after surgery to improve surgical outcomes and skills. However, surgical data is scarce, heterogeneous, and complex, which limits the applicability of existing machine learning methods. In this work, we introduce the novel task of future video generation in laparoscopic surgery. This task can augment and enrich the existing surgical data and enable various applications, such as simulation, analysis, and robot-aided surgery. Ultimately, it involves not only understanding the current state of the operation but also accurately predicting the dynamic and often unpredictable nature of surgical procedures. Our proposed method, VISAGE (VIdeo Synthesis using Action Graphs for Surgery), leverages the power of action scene graphs to capture the sequential nature of laparoscopic procedures and utilizes diffusion models to synthesize temporally coherent video sequences. VISAGE predicts the future frames given only a single initial frame, and the action graph triplets. By incorporating domain-specific knowledge through the action graph, VISAGE ensures the generated videos adhere to the expected visual and motion patterns observed in real laparoscopic procedures. The results of our experiments demonstrate high-fidelity video generation for laparoscopy procedures, which enables various applications in SDS.

comment: Accepted at MICCAI 2024 Embodied AI and Robotics for HealTHcare (EARTH) Workshop

☆ Can Uncertainty Quantification Enable Better Learning-based Index Tuning?

Tao Yu, Zhaonian Zou, Hao Xiong

Index tuning is crucial for optimizing database performance by selecting optimal indexes based on workload. The key to this process lies in an accurate and efficient benefit estimator. Traditional methods relying on what-if tools often suffer from inefficiency and inaccuracy. In contrast, learning-based models provide a promising alternative but face challenges such as instability, lack of interpretability, and complex management. To overcome these limitations, we adopt a novel approach: quantifying the uncertainty in learning-based models' results, thereby combining the strengths of both traditional and learning-based methods for reliable index tuning. We propose Beauty, the first uncertainty-aware framework that enhances learning-based models with uncertainty quantification and uses what-if tools as a complementary mechanism to improve reliability and reduce management complexity. Specifically, we introduce a novel method that combines AutoEncoder and Monte Carlo Dropout to jointly quantify uncertainty, tailored to the characteristics of benefit estimation tasks. In experiments involving sixteen models, our approach outperformed existing uncertainty quantification methods in the majority of cases. We also conducted index tuning tests on six datasets. By applying the Beauty framework, we eliminated worst-case scenarios and more than tripled the occurrence of best-case scenarios.

comment: 14 pages, 11 figures

☆ Learning Versatile Skills with Curriculum Masking NeurIPS 2024

Yao Tang, Zhihui Xie, Zichuan Lin, Deheng Ye, Shuai Li

Masked prediction has emerged as a promising pretraining paradigm in offline reinforcement learning (RL) due to its versatile masking schemes, enabling flexible inference across various downstream tasks with a unified model. Despite the versatility of masked prediction, it remains unclear how to balance the learning of skills at different levels of complexity. To address this, we propose CurrMask, a curriculum masking pretraining paradigm for sequential decision making. Motivated by how humans learn by organizing knowledge in a curriculum, CurrMask adjusts its masking scheme during pretraining for learning versatile skills. Through extensive experiments, we show that CurrMask exhibits superior zero-shot performance on skill prompting tasks, goal-conditioned planning tasks, and competitive finetuning performance on offline RL tasks. Additionally, our analysis of training dynamics reveals that CurrMask gradually acquires skills of varying complexity by dynamically adjusting its masking scheme.

comment: NeurIPS 2024 poster, 21 pages, 7 figures

☆ Continual Learning on a Data Diet

Elif Ceren Gok Yildirim, Murat Onur Yildirim, Joaquin Vanschoren

Continual Learning (CL) methods usually learn from all available data. However, this is not the case in human cognition which efficiently focuses on key experiences while disregarding the redundant information. Similarly, not all data points in a dataset have equal potential; some can be more informative than others. This disparity may significantly impact the performance, as both the quality and quantity of samples directly influence the model's generalizability and efficiency. Drawing inspiration from this, we explore the potential of learning from important samples and present an empirical study for evaluating coreset selection techniques in the context of CL to stimulate research in this unexplored area. We train different continual learners on increasing amounts of selected samples and investigate the learning-forgetting dynamics by shedding light on the underlying mechanisms driving their improved stability-plasticity balance. We present several significant observations: learning from selectively chosen samples (i) enhances incremental accuracy, (ii) improves knowledge retention of previous tasks, and (iii) refines learned representations. This analysis contributes to a deeper understanding of selective learning strategies in CL scenarios.

comment: 18 pages, 6 figures

☆ Beware of Calibration Data for Pruning Large Language Models

Yixin Ji, Yang Xiang, Juntao Li, Qingrong Xia, Ping Li, Xinyu Duan, Zhefeng Wang, Min Zhang

As large language models (LLMs) are widely applied across various fields, model compression has become increasingly crucial for reducing costs and improving inference efficiency. Post-training pruning is a promising method that does not require resource-intensive iterative training and only needs a small amount of calibration data to assess the importance of parameters. Previous research has primarily focused on designing advanced pruning methods, while different calibration data's impact on pruning performance still lacks systematical exploration. We fill this blank and surprisingly observe that the effects of calibration data even value more than designing advanced pruning strategies, especially for high sparsity. Our preliminary exploration also discloses that using calibration data similar to the training data can yield better performance. As pre-training data is usually inaccessible for advanced LLMs, we further provide a self-generating calibration data synthesis strategy to construct feasible calibration data. We conduct experiments on the recent strong open-source LLMs (e.g., DCLM, and LLaMA-3), and the results show that the proposed method outperforms commonly used calibration data and can effectively enhance strong pruning methods (e.g., Wanda, OWL).

comment: under review

☆ Scalable Random Feature Latent Variable Models

Ying Li, Zhidi Lin, Yuhao Liu, Michael Minyi Zhang, Pablo M. Olmos, Petar M. Djurić

Random feature latent variable models (RFLVMs) represent the state-of-the-art in latent variable models, capable of handling non-Gaussian likelihoods and effectively uncovering patterns in high-dimensional data. However, their heavy reliance on Monte Carlo sampling results in scalability issues which makes it difficult to use these models for datasets with a massive number of observations. To scale up RFLVMs, we turn to the optimization-based variational Bayesian inference (VBI) algorithm which is known for its scalability compared to sampling-based methods. However, implementing VBI for RFLVMs poses challenges, such as the lack of explicit probability distribution functions (PDFs) for the Dirichlet process (DP) in the kernel learning component, and the incompatibility of existing VBI algorithms with RFLVMs. To address these issues, we introduce a stick-breaking construction for DP to obtain an explicit PDF and a novel VBI algorithm called ``block coordinate descent variational inference" (BCD-VI). This enables the development of a scalable version of RFLVMs, or in short, SRFLVM. Our proposed method shows scalability, computational efficiency, superior performance in generating informative latent representations and the ability of imputing missing data across various real-world datasets, outperforming state-of-the-art competitors.

☆ Optimizing Load Scheduling in Power Grids Using Reinforcement Learning and Markov Decision Processes

Dongwen Luo

Power grid load scheduling is a critical task that ensures the balance between electricity generation and consumption while minimizing operational costs and maintaining grid stability. Traditional optimization methods often struggle with the dynamic and stochastic nature of power systems, especially when faced with renewable energy sources and fluctuating demand. This paper proposes a reinforcement learning (RL) approach using a Markov Decision Process (MDP) framework to address the challenges of dynamic load scheduling. The MDP is defined by a state space representing grid conditions, an action space covering control operations like generator adjustments and storage management, and a reward function balancing economic efficiency and system reliability. We investigate the application of various RL algorithms, from basic Q-Learning to more advanced Deep Q-Networks (DQN) and Actor-Critic methods, to determine optimal scheduling policies. The proposed approach is evaluated through a simulated power grid environment, demonstrating its potential to improve scheduling efficiency and adapt to variable demand patterns. Our results show that the RL-based method provides a robust and scalable solution for real-time load scheduling, contributing to the efficient management of modern power grids.

☆ PETAH: Parameter Efficient Task Adaptation for Hybrid Transformers in a resource-limited Context

Maximilian Augustin, Syed Shakib Sarwar, Mostafa Elhoushi, Sai Qian Zhang, Yuecheng Li, Barbara De Salvo

Following their success in natural language processing (NLP), there has been a shift towards transformer models in computer vision. While transformers perform well and offer promising multi-tasking performance, due to their high compute requirements, many resource-constrained applications still rely on convolutional or hybrid models that combine the benefits of convolution and attention layers and achieve the best results in the sub 100M parameter range. Simultaneously, task adaptation techniques that allow for the use of one shared transformer backbone for multiple downstream tasks, resulting in great storage savings at negligible cost in performance, have not yet been adopted for hybrid transformers. In this work, we investigate how to achieve the best task-adaptation performance and introduce PETAH: Parameter Efficient Task Adaptation for Hybrid Transformers. We further combine PETAH adaptation with pruning to achieve highly performant and storage friendly models for multi-tasking. In our extensive evaluation on classification and other vision tasks, we demonstrate that our PETAH-adapted hybrid models outperform established task-adaptation techniques for ViTs while requiring fewer parameters and being more efficient on mobile hardware.

☆ Mapping the Media Landscape: Predicting Factual Reporting and Political Bias Through Web Interactions

Dairazalia Sánchez-Cortés, Sergio Burdisso, Esaú Villatoro-Tello, Petr Motlicek

Bias assessment of news sources is paramount for professionals, organizations, and researchers who rely on truthful evidence for information gathering and reporting. While certain bias indicators are discernible from content analysis, descriptors like political bias and fake news pose greater challenges. In this paper, we propose an extension to a recently presented news media reliability estimation method that focuses on modeling outlets and their longitudinal web interactions. Concretely, we assess the classification performance of four reinforcement learning strategies on a large news media hyperlink graph. Our experiments, targeting two challenging bias descriptors, factual reporting and political bias, showed a significant performance improvement at the source media level. Additionally, we validate our methods on the CLEF 2023 CheckThat! Lab challenge, outperforming the reported results in both, F1-score and the official MAE metric. Furthermore, we contribute by releasing the largest annotated dataset of news source media, categorized with factual reporting and political bias labels. Our findings suggest that profiling news media sources based on their hyperlink interactions over time is feasible, offering a bird's-eye view of evolving media landscapes.

comment: Accepted to CLEF 2024

☆ Towards Active Participant-Centric Vertical Federated Learning: Some Representations May Be All You Need

Jon Irureta, Jon Imaz, Aizea Lojo, Marco González, Iñigo Perona

Vertical Federated Learning (VFL) enables collaborative model training across different participants with distinct features and common samples, while preserving data privacy. Existing VFL methodologies often struggle with realistic data partitions, typically incurring high communication costs and significant operational complexity. In this work, we introduce a novel simplified approach to VFL, Active Participant-Centric VFL (APC-VFL), that, to the best of our knowledge, is the first to require only a single communication round between participants, and allows the active participant to do inference in a non collaborative fashion. This method integrates unsupervised representation learning with knowledge distillation to achieve comparable accuracy to traditional VFL methods based on vertical split learning in classical settings, reducing required communication rounds by up to $4200\times$, while being more flexible. Our approach also shows improvements compared to non-federated local models, as well as a comparable VFL proposal, VFedTrans, offering an efficient and flexible solution for collaborative learning.

☆ Entity-based Reinforcement Learning for Autonomous Cyber Defence CCS 2024

Isaac Symes Thompson, Alberto Caron, Chris Hicks, Vasilios Mavroudis

A significant challenge for autonomous cyber defence is ensuring a defensive agent's ability to generalise across diverse network topologies and configurations. This capability is necessary for agents to remain effective when deployed in dynamically changing environments, such as an enterprise network where devices may frequently join and leave. Standard approaches to deep reinforcement learning, where policies are parameterised using a fixed-input multi-layer perceptron (MLP) expect fixed-size observation and action spaces. In autonomous cyber defence, this makes it hard to develop agents that generalise to environments with network topologies different from those trained on, as the number of nodes affects the natural size of the observation and action spaces. To overcome this limitation, we reframe the problem of autonomous network defence using entity-based reinforcement learning, where the observation and action space of an agent are decomposed into a collection of discrete entities. This framework enables the use of policy parameterisations specialised in compositional generalisation. Namely, we train a Transformer-based policy on the Yawning Titan cyber-security simulation environment and test its generalisation capabilities across various network topologies. We demonstrate that this approach significantly outperforms an MLP-based policy on fixed networks, and has the ability for zero-shot generalisation to networks of a different size to those seen in training. These findings highlight the potential for entity-based reinforcement learning to advance the field of autonomous cyber defence by providing more generalisable policies capable of handling variations in real-world network environments.

comment: Material to appear in the proceedings of the 1st International Workshop on Autonomous Cybersecurity at ACM CCS 2024

☆ Exploring structure diversity in atomic resolution microscopy with graph neural networks

Zheng Luo, Ming Feng, Zijian Gao, Jinyang Yu, Liang Hu, Tao Wang, Shenao Xue, Shen Zhou, Fangping Ouyang, Dawei Feng, Kele Xu, Shanshan Wang

The emergence of deep learning (DL) has provided great opportunities for the high-throughput analysis of atomic-resolution micrographs. However, the DL models trained by image patches in fixed size generally lack efficiency and flexibility when processing micrographs containing diversified atomic configurations. Herein, inspired by the similarity between the atomic structures and graphs, we describe a few-shot learning framework based on an equivariant graph neural network (EGNN) to analyze a library of atomic structures (e.g., vacancies, phases, grain boundaries, doping, etc.), showing significantly promoted robustness and three orders of magnitude reduced computing parameters compared to the image-driven DL models, which is especially evident for those aggregated vacancy lines with flexible lattice distortion. Besides, the intuitiveness of graphs enables quantitative and straightforward extraction of the atomic-scale structural features in batches, thus statistically unveiling the self-assembly dynamics of vacancy lines under electron beam irradiation. A versatile model toolkit is established by integrating EGNN sub-models for single structure recognition to process images involving varied configurations in the form of a task chain, leading to the discovery of novel doping configurations with superior electrocatalytic properties for hydrogen evolution reactions. This work provides a powerful tool to explore structure diversity in a fast, accurate, and intelligent manner.

☆ Feature Learning in Attention Mechanisms Is More Compact and Stable Than in Convolution

Baiyuan Chen

Attention and convolution are fundamental techniques in machine learning. While they use different approaches to learn features - attention mechanisms capture both global and local data relathionships, while convolutional layers focus on local patterns - both methods are effective for various tasks. Although the feature learning of both models is well-studied individually, there has not been a direct comparison of their feature learning dynamics. In this paper, we compare their Lipschitz continuity with respect to the Wasserstein distance and covering numbers under similar settings. We demonstrate that attention processes data in a more compact and stable manner. Compactness refers to the lower variance and intrinsic dimensionality of the activation outputs, while stability refers to the changes between inputs and outputs. We validate our findings through experiments using topological data analysis, measuring the 1-, 2-, and infinity-Wasserstein distances between the outputs of each layer from both models. Furthermore, we extend our comparison to Vision Transformers (ViTs) and ResNets, showing that while ViTs have higher output variance, their feature learning is more stable than that of ResNets.

☆ Incremental Learning of Affordances using Markov Logic Networks

George Potter, Gertjan Burghouts, Joris Sijs

Affordances enable robots to have a semantic understanding of their surroundings. This allows them to have more acting flexibility when completing a given task. Capturing object affordances in a machine learning model is a difficult task, because of their dependence on contextual information. Markov Logic Networks (MLN) combine probabilistic reasoning with logic that is able to capture such context. Mobile robots operate in partially known environments wherein unseen object affordances can be observed. This new information must be incorporated into the existing knowledge, without having to retrain the MLN from scratch. We introduce the MLN Cumulative Learning Algorithm (MLN-CLA). MLN-CLA learns new relations in various knowledge domains by retaining knowledge and only updating the changed knowledge, for which the MLN is retrained. We show that MLN-CLA is effective for accumulative learning and zero-shot affordance inference, outperforming strong baselines.

comment: accepted at IEEE IRC 2024

☆ Self-Supervised Graph Neural Networks for Enhanced Feature Extraction in Heterogeneous Information Networks

Jianjun Wei, Yue Liu, Xin Huang, Xin Zhang, Wenyi Liu, Xu Yan

This paper explores the applications and challenges of graph neural networks (GNNs) in processing complex graph data brought about by the rapid development of the Internet. Given the heterogeneity and redundancy problems that graph data often have, traditional GNN methods may be overly dependent on the initial structure and attribute information of the graph, which limits their ability to accurately simulate more complex relationships and patterns in the graph. Therefore, this study proposes a graph neural network model under a self-supervised learning framework, which can flexibly combine different types of additional information of the attribute graph and its nodes, so as to better mine the deep features in the graph data. By introducing a self-supervisory mechanism, it is expected to improve the adaptability of existing models to the diversity and complexity of graph data and improve the overall performance of the model.

☆ A Kernel Perspective on Distillation-based Collaborative Learning NeurIPS 2024

Sejun Park, Kihun Hong, Ganguk Hwang

Over the past decade, there is a growing interest in collaborative learning that can enhance AI models of multiple parties. However, it is still challenging to enhance performance them without sharing private data and models from individual parties. One recent promising approach is to develop distillation-based algorithms that exploit unlabeled public data but the results are still unsatisfactory in both theory and practice. To tackle this problem, we rigorously analyze a representative distillation-based algorithm in the view of kernel regression. This work provides the first theoretical results to prove the (nearly) minimax optimality of the nonparametric collaborative learning algorithm that does not directly share local data or models in massively distributed statistically heterogeneous environments. Inspired by our theoretical results, we also propose a practical distillation-based collaborative learning algorithm based on neural network architecture. Our algorithm successfully bridges the gap between our theoretical assumptions and practical settings with neural networks through feature kernel matching. We simulate various regression tasks to verify our theory and demonstrate the practical feasibility of our proposed algorithm.

comment: Accepted to NeurIPS 2024

☆ Challenge on Sound Scene Synthesis: Evaluating Text-to-Audio Generation NeurIPS 2024

Junwon Lee, Modan Tailleur, Laurie M. Heller, Keunwoo Choi, Mathieu Lagrange, Brian McFee, Keisuke Imoto, Yuki Okamoto

Despite significant advancements in neural text-to-audio generation, challenges persist in controllability and evaluation. This paper addresses these issues through the Sound Scene Synthesis challenge held as part of the Detection and Classification of Acoustic Scenes and Events 2024. We present an evaluation protocol combining objective metric, namely Fr\'echet Audio Distance, with perceptual assessments, utilizing a structured prompt format to enable diverse captions and effective evaluation. Our analysis reveals varying performance across sound categories and model architectures, with larger models generally excelling but innovative lightweight approaches also showing promise. The strong correlation between objective metrics and human ratings validates our evaluation approach. We discuss outcomes in terms of audio quality, controllability, and architectural considerations for text-to-audio synthesizers, providing direction for future research.

comment: accepted to NeurIPS 2024 Workshop: Audio Imagination

☆ Predicting Company Growth by Econophysics informed Machine Learning

Ruyi Tao, Kaiwei Liu, Xu Jing, Jiang Zhang

Predicting company growth is crucial for strategic adjustment, operational decision-making, risk assessment, and loan eligibility reviews. Traditional models for company growth often focus too much on theory, overlooking practical forecasting, or they rely solely on time series forecasting techniques, ignoring interpretability and the inherent mechanisms of company growth. In this paper, we propose a machine learning-based prediction framework that incorporates an econophysics model for company growth. Our model captures both the intrinsic growth mechanisms of companies led by scaling laws and the fluctuations influenced by random factors and individual decisions, demonstrating superior predictive performance compared with methods that use time series techniques alone. Its advantages are more pronounced in long-range prediction tasks. By explicitly modeling the baseline growth and volatility components, our model is more interpretable.

comment: 18 pages, 12 figures

☆ Bonsai: Gradient-free Graph Distillation for Node Classification

Mridul Gupta, Samyak Jain, Vansh Ramani, Hariprasad Kodamana, Sayan Ranu

Graph distillation has emerged as a promising avenue to enable scalable training of GNNs by compressing the training dataset while preserving essential graph characteristics. Our study uncovers significant shortcomings in current graph distillation techniques. First, the majority of the algorithms paradoxically require training on the full dataset to perform distillation. Second, due to their gradient-emulating approach, these methods require fresh distillation for any change in hyperparameters or GNN architecture, limiting their flexibility and reusability. Finally, they fail to achieve substantial size reduction due to synthesizing fully-connected, edge-weighted graphs. To address these challenges, we present Bonsai, a novel graph distillation method empowered by the observation that \textit{computation trees} form the fundamental processing units of message-passing GNNs. Bonsai distills datasets by encoding a careful selection of \textit{exemplar} trees that maximize the representation of all computation trees in the training set. This unique approach imparts Bonsai as the first linear-time, model-agnostic graph distillation algorithm for node classification that outperforms existing baselines across $6$ real-world datasets on accuracy, while being $22$ times faster on average. Bonsai is grounded in rigorous mathematical guarantees on the adopted approximation strategies making it robust to GNN architectures, datasets, and parameters.

☆ Adversarial Domain Adaptation for Metal Cutting Sound Detection: Leveraging Abundant Lab Data for Scarce Industry Data

Mir Imtiaz Mostafiz, Eunseob Kim, Adrian Shuai Li, Elisa Bertino, Martin Byung-Guk Jun, Ali Shakouri

Cutting state monitoring in the milling process is crucial for improving manufacturing efficiency and tool life. Cutting sound detection using machine learning (ML) models, inspired by experienced machinists, can be employed as a cost-effective and non-intrusive monitoring method in a complex manufacturing environment. However, labeling industry data for training is costly and time-consuming. Moreover, industry data is often scarce. In this study, we propose a novel adversarial domain adaptation (DA) approach to leverage abundant lab data to learn from scarce industry data, both labeled, for training a cutting-sound detection model. Rather than adapting the features from separate domains directly, we project them first into two separate latent spaces that jointly work as the feature space for learning domain-independent representations. We also analyze two different mechanisms for adversarial learning where the discriminator works as an adversary and a critic in separate settings, enabling our model to learn expressive domain-invariant and domain-ingrained features, respectively. We collected cutting sound data from multiple sensors in different locations, prepared datasets from lab and industry domain, and evaluated our learning models on them. Experiments showed that our models outperformed the multi-layer perceptron based vanilla domain adaptation models in labeling tasks on the curated datasets, achieving near 92%, 82% and 85% accuracy respectively for three different sensors installed in industry settings.

comment: 8 pages, 3 figures, 3 tables, First two named Authors have equal contribution (Co-first author)

☆ Securing Federated Learning Against Novel and Classic Backdoor Threats During Foundation Model Integration

Xiaohuan Bi, Xi Li

Federated learning (FL) enables decentralized model training while preserving privacy. Recently, integrating Foundation Models (FMs) into FL has boosted performance but also introduced a novel backdoor attack mechanism. Attackers can exploit the FM's capabilities to embed backdoors into synthetic data generated by FMs used for model fusion, subsequently infecting all client models through knowledge sharing without involvement in the long-lasting FL process. These novel attacks render existing FL backdoor defenses ineffective, as they primarily detect anomalies among client updates, which may appear uniformly malicious under this attack. Our work proposes a novel data-free defense strategy by constraining abnormal activations in the hidden feature space during model aggregation on the server. The activation constraints, optimized using synthetic data alongside FL training, mitigate the attack while barely affecting model performance, as the parameters remain untouched. Extensive experiments demonstrate its effectiveness against both novel and classic backdoor attacks, outperforming existing defenses while maintaining model performance.

☆ Differentially Private Learning Needs Better Model Initialization and Self-Distillation

Ivoline C. Ngong, Joseph P. Near, Niloofar Mireshghallah

Differentially private SGD (DPSGD) enables privacy-preserving training of language models, but often reduces utility, diversity, and linguistic quality. We introduce DPRefine, a three-phase method that initializes a model using data synthesis from a small pre-trained LM with rigorous filtering, applies DP finetuning on private data, and performs self-distillation to refine outputs. This approach significantly outperforms vanilla DPSGD, with AlpacaEval preferring DPRefine's generations in 78.4% of cases across all datasets. Our analysis reveals that DPRefine reduces linguistic errors in generated text by 84.0%, mitigating grammar and spelling errors, commonly associated with DPSGD. It also reduces inconsistencies of non-private models, such as hallucinated details and misattributed quotes. We find that small models like GPT-2 can be effective for initialization and distillation, highlighting their potential in enabling scalable and efficient deployment of privacy-preserving language.

comment: 18 pages

☆ DisenGCD: A Meta Multigraph-assisted Disentangled Graph Learning Framework for Cognitive Diagnosis NeurIPS 2024

Shangshang Yang, Mingyang Chen, Ziwen Wang, Xiaoshan Yu, Panpan Zhang, Haiping Ma, Xingyi Zhang

Existing graph learning-based cognitive diagnosis (CD) methods have made relatively good results, but their student, exercise, and concept representations are learned and exchanged in an implicit unified graph, which makes the interaction-agnostic exercise and concept representations be learned poorly, failing to provide high robustness against noise in students' interactions. Besides, lower-order exercise latent representations obtained in shallow layers are not well explored when learning the student representation. To tackle the issues, this paper suggests a meta multigraph-assisted disentangled graph learning framework for CD (DisenGCD), which learns three types of representations on three disentangled graphs: student-exercise-concept interaction, exercise-concept relation, and concept dependency graphs, respectively. Specifically, the latter two graphs are first disentangled from the interaction graph. Then, the student representation is learned from the interaction graph by a devised meta multigraph learning module; multiple learnable propagation paths in this module enable current student latent representation to access lower-order exercise latent representations, which can lead to more effective nad robust student representations learned; the exercise and concept representations are learned on the relation and dependency graphs by graph attention modules. Finally, a novel diagnostic function is devised to handle three disentangled representations for prediction. Experiments show better performance and robustness of DisenGCD than state-of-the-art CD methods and demonstrate the effectiveness of the disentangled learning framework and meta multigraph module. The source code is available at \textcolor{red}{\url{https://github.com/BIMK/Intelligent-Education/tree/main/DisenGCD}}.

comment: 21 pages, Accepted by NeurIPS 2024 as a poster

☆ BlurryScope: a cost-effective and compact scanning microscope for automated HER2 scoring using deep learning on blurry image data

Michael John Fanous, Christopher Michael Seybold, Hanlong Chen, Nir Pillar, Aydogan Ozcan

We developed a rapid scanning optical microscope, termed "BlurryScope", that leverages continuous image acquisition and deep learning to provide a cost-effective and compact solution for automated inspection and analysis of tissue sections. BlurryScope integrates specialized hardware with a neural network-based model to quickly process motion-blurred histological images and perform automated pathology classification. This device offers comparable speed to commercial digital pathology scanners, but at a significantly lower price point and smaller size/weight, making it ideal for fast triaging in small clinics, as well as for resource-limited settings. To demonstrate the proof-of-concept of BlurryScope, we implemented automated classification of human epidermal growth factor receptor 2 (HER2) scores on immunohistochemically (IHC) stained breast tissue sections, achieving concordant results with those obtained from a high-end digital scanning microscope. We evaluated this approach by scanning HER2-stained tissue microarrays (TMAs) at a continuous speed of 5 mm/s, which introduces bidirectional motion blur artifacts. These compromised images were then used to train our network models. Using a test set of 284 unique patient cores, we achieved blind testing accuracies of 79.3% and 89.7% for 4-class (0, 1+, 2+, 3+) and 2-class (0/1+ , 2+/3+) HER2 score classification, respectively. BlurryScope automates the entire workflow, from image scanning to stitching and cropping of regions of interest, as well as HER2 score classification. We believe BlurryScope has the potential to enhance the current pathology infrastructure in resource-scarce environments, save diagnostician time and bolster cancer identification and classification across various clinical environments.

comment: 18 Pages, 6 Figures

☆ LEADS: Lightweight Embedded Assisted Driving System

Tianhao Fu, Querobin Mascarenhas, Andrew Forti

With the rapid development of electric vehicles, formula races that face high school and university students have become more popular than ever as the threshold for design and manufacturing has been lowered. In many cases, we see teams inspired by or directly using toolkits and technologies inherited from standardized commercial vehicles. These architectures are usually overly complicated for amateur applications like the races. In order to improve the efficiency and simplify the development of instrumentation, control, and analysis systems, we propose LEADS (Lightweight Embedded Assisted Driving System), a dedicated solution for such scenarios.

comment: 12 pages

☆ Multimodal Information Bottleneck for Deep Reinforcement Learning with Multiple Sensors

Bang You, Huaping Liu

Reinforcement learning has achieved promising results on robotic control tasks but struggles to leverage information effectively from multiple sensory modalities that differ in many characteristics. Recent works construct auxiliary losses based on reconstruction or mutual information to extract joint representations from multiple sensory inputs to improve the sample efficiency and performance of reinforcement learning algorithms. However, the representations learned by these methods could capture information irrelevant to learning a policy and may degrade the performance. We argue that compressing information in the learned joint representations about raw multimodal observations is helpful, and propose a multimodal information bottleneck model to learn task-relevant joint representations from egocentric images and proprioception. Our model compresses and retains the predictive information in multimodal observations for learning a compressed joint representation, which fuses complementary information from visual and proprioceptive feedback and meanwhile filters out task-irrelevant information in raw multimodal observations. We propose to minimize the upper bound of our multimodal information bottleneck objective for computationally tractable optimization. Experimental evaluations on several challenging locomotion tasks with egocentric images and proprioception show that our method achieves better sample efficiency and zero-shot robustness to unseen white noise than leading baselines. We also empirically demonstrate that leveraging information from egocentric images and proprioception is more helpful for learning policies on locomotion tasks than solely using one single modality.

comment: 31 pages

☆ Predicting 30-Day Hospital Readmission in Medicare Patients: Insights from an LSTM Deep Learning Model

Xintao Li, Sibei Liu, Dezhi Yu, Yang Zhang, Xiaoyu Liu

Readmissions among Medicare beneficiaries are a major problem for the US healthcare system from a perspective of both healthcare operations and patient caregiving outcomes. Our study analyzes Medicare hospital readmissions using LSTM networks with feature engineering to assess feature contributions. We selected variables from admission-level data, inpatient medical history and patient demography. The LSTM model is designed to capture temporal dynamics from admission-level and patient-level data. On a case study on the MIMIC dataset, the LSTM model outperformed the logistic regression baseline, accurately leveraging temporal features to predict readmission. The major features were the Charlson Comorbidity Index, hospital length of stay, the hospital admissions over the past 6 months, while demographic variables were less impactful. This work suggests that LSTM networks offers a more promising approach to improve Medicare patient readmission prediction. It captures temporal interactions in patient databases, enhancing current prediction models for healthcare providers. Adoption of predictive models into clinical practice may be more effective in identifying Medicare patients to provide early and targeted interventions to improve patient outcomes.

comment: 5 pages, 1 table, 5 figures, Accepted by 2024 3rd International Conference on Cloud Computing, Big Data Application and Software Engineering(CBASE 2024), the final version will be published on on IEEE Conference proceeding

☆ Primal-Dual Spectral Representation for Off-policy Evaluation

Yang Hu, Tianyi Chen, Na Li, Kai Wang, Bo Dai

Off-policy evaluation (OPE) is one of the most fundamental problems in reinforcement learning (RL) to estimate the expected long-term payoff of a given target policy with only experiences from another behavior policy that is potentially unknown. The distribution correction estimation (DICE) family of estimators have advanced the state of the art in OPE by breaking the curse of horizon. However, the major bottleneck of applying DICE estimators lies in the difficulty of solving the saddle-point optimization involved, especially with neural network implementations. In this paper, we tackle this challenge by establishing a linear representation of value function and stationary distribution correction ratio, i.e., primal and dual variables in the DICE framework, using the spectral decomposition of the transition operator. Such primal-dual representation not only bypasses the non-convex non-concave optimization in vanilla DICE, therefore enabling an computational efficient algorithm, but also paves the way for more efficient utilization of historical data. We highlight that our algorithm, SpectralDICE, is the first to leverage the linear representation of primal-dual variables that is both computation and sample efficient, the performance of which is supported by a rigorous theoretical sample complexity guarantee and a thorough empirical evaluation on various benchmarks.

comment: 29 pages, 5 figures

☆ GDDA: Semantic OOD Detection on Graphs under Covariate Shift via Score-Based Diffusion Models

Zhixia He, Chen Zhao, Minglai Shao, Yujie Lin, Dong Li, Qin Tian

Out-of-distribution (OOD) detection poses a significant challenge for Graph Neural Networks (GNNs), particularly in open-world scenarios with varying distribution shifts. Most existing OOD detection methods on graphs primarily focus on identifying instances in test data domains caused by either semantic shifts (changes in data classes) or covariate shifts (changes in data features), while leaving the simultaneous occurrence of both distribution shifts under-explored. In this work, we address both types of shifts simultaneously and introduce a novel challenge for OOD detection on graphs: graph-level semantic OOD detection under covariate shift. In this scenario, variations between the training and test domains result from the concurrent presence of both covariate and semantic shifts, where only graphs associated with unknown classes are identified as OOD samples (OODs). To tackle this challenge, we propose a novel two-phase framework called Graph Disentangled Diffusion Augmentation (GDDA). The first phase focuses on disentangling graph representations into domain-invariant semantic factors and domain-specific style factors. In the second phase, we introduce a novel distribution-shift-controlled score-based generative diffusion model that generates latent factors outside the training semantic and style spaces. Additionally, auxiliary pseudo-in-distribution (InD) and pseudo-OOD graph representations are employed to enhance the effectiveness of the energy-based semantic OOD detector. Extensive empirical studies on three benchmark datasets demonstrate that our approach outperforms state-of-the-art baselines.

comment: 4 pages, 6 figures

☆ MobileSafetyBench: Evaluating Safety of Autonomous Agents in Mobile Device Control

Juyong Lee, Dongyoon Hahm, June Suk Choi, W. Bradley Knox, Kimin Lee

Autonomous agents powered by large language models (LLMs) show promising potential in assistive tasks across various domains, including mobile device control. As these agents interact directly with personal information and device settings, ensuring their safe and reliable behavior is crucial to prevent undesirable outcomes. However, no benchmark exists for standardized evaluation of the safety of mobile device-control agents. In this work, we introduce MobileSafetyBench, a benchmark designed to evaluate the safety of device-control agents within a realistic mobile environment based on Android emulators. We develop a diverse set of tasks involving interactions with various mobile applications, including messaging and banking applications. To clearly evaluate safety apart from general capabilities, we design separate tasks measuring safety and tasks evaluating helpfulness. The safety tasks challenge agents with managing potential risks prevalent in daily life and include tests to evaluate robustness against indirect prompt injections. Our experiments demonstrate that while baseline agents, based on state-of-the-art LLMs, perform well in executing helpful tasks, they show poor performance in safety tasks. To mitigate these safety concerns, we propose a prompting method that encourages agents to prioritize safety considerations. While this method shows promise in promoting safer behaviors, there is still considerable room for improvement to fully earn user trust. This highlights the urgent need for continued research to develop more robust safety mechanisms in mobile environments. We open-source our benchmark at: https://mobilesafetybench.github.io/.

☆ Univariate Conditional Variational Autoencoder for Morphogenic Patterns Design in Frontal Polymerization-Based Manufacturing

Qibang Liu, Pengfei Cai, Diab Abueidda, Seid Koric, Rafael Gomez-Bombarellig, Philippe Geubelle

Rapid reaction-thermal diffusion during frontal polymerization (FP) with variations in initial and boundary conditions destabilizes the planar mode of front propagation, leading to spatially varying complex hierarchical patterns in polymeric materials. Although modern reaction-diffusion models can predict the patterns resulting from unstable FP, the inverse design of patterns, which aims to retrieve process conditions that produce a desired pattern, remains an open challenge due to the nonunique and nonintuitive mapping between process conditions and patterns. In this work, we propose a novel probabilistic generative model named univariate conditional variational autoencoder (UcVAE) for the inverse design of hierarchical patterns in FP-based manufacturing. Unlike the cVAE, which encodes both the design space and the design target, the UcVAE encodes only the design space. In the encoder of the UcVAE, the number of training parameters is significantly reduced compared to the cVAE, resulting in a shorter training time while maintaining comparable performance. Given desired pattern images, the trained UcVAE can generate multiple process condition solutions that produce high-fidelity hierarchical patterns.

☆ Time and Frequency Synergy for Source-Free Time-Series Domain Adaptations

Muhammad Tanzil Furqon, Mahardhika Pratama, Ary Mazharuddin Shiddiqi, Lin Liu, Habibullah Habibullah, Kutluyil Dogancay

The issue of source-free time-series domain adaptations still gains scarce research attentions. On the other hand, existing approaches rely solely on time-domain features ignoring frequency components providing complementary information. This paper proposes Time Frequency Domain Adaptation (TFDA), a method to cope with the source-free time-series domain adaptation problems. TFDA is developed with a dual branch network structure fully utilizing both time and frequency features in delivering final predictions. It induces pseudo-labels based on a neighborhood concept where predictions of a sample group are aggregated to generate reliable pseudo labels. The concept of contrastive learning is carried out in both time and frequency domains with pseudo label information and a negative pair exclusion strategy to make valid neighborhood assumptions. In addition, the time-frequency consistency technique is proposed using the self-distillation strategy while the uncertainty reduction strategy is implemented to alleviate uncertainties due to the domain shift problem. Last but not least, the curriculum learning strategy is integrated to combat noisy pseudo labels. Our experiments demonstrate the advantage of our approach over prior arts with noticeable margins in benchmark problems.

☆ Congestion Forecast for Trains with Railroad-Graph-based Semi-Supervised Learning using Sparse Passenger Reports SP

Soto Anno, Kota Tsubouchi, Masamichi Shimosaka

Forecasting rail congestion is crucial for efficient mobility in transport systems. We present rail congestion forecasting using reports from passengers collected through a transit application. Although reports from passengers have received attention from researchers, ensuring a sufficient volume of reports is challenging due to passenger's reluctance. The limited number of reports results in the sparsity of the congestion label, which can be an issue in building a stable prediction model. To address this issue, we propose a semi-supervised method for congestion forecasting for trains, or SURCONFORT. Our key idea is twofold: firstly, we adopt semi-supervised learning to leverage sparsely labeled data and many unlabeled data. Secondly, in order to complement the unlabeled data from nearby stations, we design a railway network-oriented graph and apply the graph to semi-supervised graph regularization. Empirical experiments with actual reporting data show that SURCONFORT improved the forecasting performance by 14.9% over state-of-the-art methods under the label sparsity.

comment: Accepted in ACM SIGSPATIAL 2024

☆ WAGLE: Strategic Weight Attribution for Effective and Modular Unlearning in Large Language Models

Jinghan Jia, Jiancheng Liu, Yihua Zhang, Parikshit Ram, Nathalie Baracaldo, Sijia Liu

The need for effective unlearning mechanisms in large language models (LLMs) is increasingly urgent, driven by the necessity to adhere to data regulations and foster ethical generative AI practices. Despite growing interest of LLM unlearning, much of the existing research has focused on varied unlearning method designs to boost effectiveness and efficiency. However, the inherent relationship between model weights and LLM unlearning has not been extensively examined. In this paper, we systematically explore how model weights interact with unlearning processes in LLMs and we design the weight attribution-guided LLM unlearning method, WAGLE, which unveils the interconnections between 'influence' of weights and 'influence' of data to forget and retain in LLM generation. By strategically guiding the LLM unlearning across different types of unlearning methods and tasks, WAGLE can erase the undesired content, while maintaining the performance of the original tasks. We refer to the weight attribution-guided LLM unlearning method as WAGLE, which unveils the interconnections between 'influence' of weights and 'influence' of data to forget and retain in LLM generation. Our extensive experiments show that WAGLE boosts unlearning performance across a range of LLM unlearning methods such as gradient difference and (negative) preference optimization, applications such as fictitious unlearning, malicious use prevention, and copyrighted information removal, and models including Zephyr-7b-beta and Llama2-7b. To the best of our knowledge, our work offers the first principled method for attributing and pinpointing the influential weights in enhancing LLM unlearning. It stands in contrast to previous methods that lack weight attribution and simpler weight attribution techniques.

☆ Mitigating Graph Covariate Shift via Score-based Out-of-distribution Augmentation

Bohan Wang, Yurui Chang, Lu Lin

Distribution shifts between training and testing datasets significantly impair the model performance on graph learning. A commonly-taken causal view in graph invariant learning suggests that stable predictive features of graphs are causally associated with labels, whereas varying environmental features lead to distribution shifts. In particular, covariate shifts caused by unseen environments in test graphs underscore the critical need for out-of-distribution (OOD) generalization. Existing graph augmentation methods designed to address the covariate shift often disentangle the stable and environmental features in the input space, and selectively perturb or mixup the environmental features. However, such perturbation-based methods heavily rely on an accurate separation of stable and environmental features, and their exploration ability is confined to existing environmental features in the training distribution. To overcome these limitations, we introduce a novel approach using score-based graph generation strategies that synthesize unseen environmental features while preserving the validity and stable features of overall graph patterns. Our comprehensive empirical evaluations demonstrate the enhanced effectiveness of our method in improving graph OOD generalization.

comment: 17 pages, 5 figures, 4 tables

☆ BadFair: Backdoored Fairness Attacks with Group-conditioned Triggers EMNLP 2024

Jiaqi Xue, Qian Lou, Mengxin Zheng

Attacking fairness is crucial because compromised models can introduce biased outcomes, undermining trust and amplifying inequalities in sensitive applications like hiring, healthcare, and law enforcement. This highlights the urgent need to understand how fairness mechanisms can be exploited and to develop defenses that ensure both fairness and robustness. We introduce BadFair, a novel backdoored fairness attack methodology. BadFair stealthily crafts a model that operates with accuracy and fairness under regular conditions but, when activated by certain triggers, discriminates and produces incorrect results for specific groups. This type of attack is particularly stealthy and dangerous, as it circumvents existing fairness detection methods, maintaining an appearance of fairness in normal use. Our findings reveal that BadFair achieves a more than 85% attack success rate in attacks aimed at target groups on average while only incurring a minimal accuracy loss. Moreover, it consistently exhibits a significant discrimination score, distinguishing between pre-defined target and non-target attacked groups across various datasets and models.

comment: Accepted by EMNLP 2024

☆ GenDP: 3D Semantic Fields for Category-Level Generalizable Diffusion Policy

Yixuan Wang, Guang Yin, Binghao Huang, Tarik Kelestemur, Jiuguang Wang, Yunzhu Li

Diffusion-based policies have shown remarkable capability in executing complex robotic manipulation tasks but lack explicit characterization of geometry and semantics, which often limits their ability to generalize to unseen objects and layouts. To enhance the generalization capabilities of Diffusion Policy, we introduce a novel framework that incorporates explicit spatial and semantic information via 3D semantic fields. We generate 3D descriptor fields from multi-view RGBD observations with large foundational vision models, then compare these descriptor fields against reference descriptors to obtain semantic fields. The proposed method explicitly considers geometry and semantics, enabling strong generalization capabilities in tasks requiring category-level generalization, resolving geometric ambiguities, and attention to subtle geometric details. We evaluate our method across eight tasks involving articulated objects and instances with varying shapes and textures from multiple object categories. Our method demonstrates its effectiveness by increasing Diffusion Policy's average success rate on unseen instances from 20% to 93%. Additionally, we provide a detailed analysis and visualization to interpret the sources of performance gain and explain how our method can generalize to novel instances.

comment: Accepted to Conference on Robot Learning (CoRL 2024). Project Page: https://robopil.github.io/GenDP/

☆ Which Client is Reliable?: A Reliable and Personalized Prompt-based Federated Learning for Medical Image Question Answering

He Zhu, Ren Togo, Takahiro Ogawa, Miki Haseyama

Conventional medical artificial intelligence (AI) models face barriers in clinical application and ethical issues owing to their inability to handle the privacy-sensitive characteristics of medical data. We present a novel personalized federated learning (pFL) method for medical visual question answering (VQA) models, addressing privacy reliability challenges in the medical domain. Our method introduces learnable prompts into a Transformer architecture to efficiently train it on diverse medical datasets without massive computational costs. Then we introduce a reliable client VQA model that incorporates Dempster-Shafer evidence theory to quantify uncertainty in predictions, enhancing the model's reliability. Furthermore, we propose a novel inter-client communication mechanism that uses maximum likelihood estimation to balance accuracy and uncertainty, fostering efficient integration of insights across clients.

♻ ☆ Understanding Transfer Learning via Mean-field Analysis

Gholamali Aminian, Łukasz Szpruch, Samuel N. Cohen

We propose a novel framework for exploring generalization errors of transfer learning through the lens of differential calculus on the space of probability measures. In particular, we consider two main transfer learning scenarios, $\alpha$-ERM and fine-tuning with the KL-regularized empirical risk minimization and establish generic conditions under which the generalization error and the population risk convergence rates for these scenarios are studied. Based on our theoretical results, we show the benefits of transfer learning with a one-hidden-layer neural network in the mean-field regime under some suitable integrability and regularity assumptions on the loss and activation functions.

comment: Under review

♻ ☆ Optimal Design for Reward Modeling in RLHF

Antoine Scheid, Etienne Boursier, Alain Durmus, Michael I. Jordan, Pierre Ménard, Eric Moulines, Michal Valko

Reinforcement Learning from Human Feedback (RLHF) has become a popular approach to align language models (LMs) with human preferences. This method involves collecting a large dataset of human pairwise preferences across various text generations and using it to infer (implicitly or explicitly) a reward model. Numerous methods have been proposed to learn the reward model and align a LM with it. However, the costly process of collecting human preferences has received little attention and could benefit from theoretical insights. This paper addresses this issue and aims to formalize the reward training model in RLHF. We frame the selection of an effective dataset as a simple regret minimization task, using a linear contextual dueling bandit method. Given the potentially large number of arms, this approach is more coherent than the best-arm identification setting. We then propose an offline framework for solving this problem. Under appropriate assumptions - linearity of the reward model in the embedding space, and boundedness of the reward parameter - we derive bounds on the simple regret. Finally, we provide a lower bound that matches our upper bound up to constant and logarithmic terms. To our knowledge, this is the first theoretical contribution in this area to provide an offline approach as well as worst-case guarantees.

♻ ☆ xLSTM-Mixer: Multivariate Time Series Forecasting by Mixing via Scalar Memories

Maurice Kraus, Felix Divo, Devendra Singh Dhami, Kristian Kersting

Time series data is prevalent across numerous fields, necessitating the development of robust and accurate forecasting models. Capturing patterns both within and between temporal and multivariate components is crucial for reliable predictions. We introduce xLSTM-Mixer, a model designed to effectively integrate temporal sequences, joint time-variate information, and multiple perspectives for robust forecasting. Our approach begins with a linear forecast shared across variates, which is then refined by xLSTM blocks. These blocks serve as key elements for modeling the complex dynamics of challenging time series data. xLSTM-Mixer ultimately reconciles two distinct views to produce the final forecast. Our extensive evaluations demonstrate xLSTM-Mixer's superior long-term forecasting performance compared to recent state-of-the-art methods. A thorough model analysis provides further insights into its key components and confirms its robustness and effectiveness. This work contributes to the resurgence of recurrent models in time series forecasting.

♻ ☆ Masked Clinical Modelling: A Framework for Synthetic and Augmented Survival Data Generation

Nicholas I-Hsien Kuo, Blanca Gallego, Louisa Jorm

Access to real clinical data is often restricted due to privacy obligations, creating significant barriers for healthcare research. Synthetic datasets provide a promising solution, enabling secure data sharing and model development. However, most existing approaches focus on data realism rather than utility -- ensuring that models trained on synthetic data yield clinically meaningful insights comparable to those trained on real data. In this paper, we present Masked Clinical Modelling (MCM), a framework inspired by masked language modelling, designed for both data synthesis and conditional data augmentation. We evaluate this prototype on the WHAS500 dataset using Cox Proportional Hazards models, focusing on the preservation of hazard ratios as key clinical metrics. Our results show that data generated using the MCM framework improves both discrimination and calibration in survival analysis, outperforming existing methods. MCM demonstrates strong potential to support survival data analysis and broader healthcare applications.

comment: Re-archived due to incorrect ORCiD. Last edited: 2024-10-23

♻ ☆ LLMScan: Causal Scan for LLM Misbehavior Detection

Mengdi Zhang, Kai Kiat Goh, Peixin Zhang, Jun Sun

Despite the success of Large Language Models (LLMs) across various fields, their potential to generate untruthful, biased and harmful responses poses significant risks, particularly in critical applications. This highlights the urgent need for systematic methods to detect and prevent such misbehavior. While existing approaches target specific issues such as harmful responses, this work introduces LLMScan, an innovative LLM monitoring technique based on causality analysis, offering a comprehensive solution. LLMScan systematically monitors the inner workings of an LLM through the lens of causal inference, operating on the premise that the LLM's `brain' behaves differently when misbehaving. By analyzing the causal contributions of the LLM's input tokens and transformer layers, LLMScan effectively detects misbehavior. Extensive experiments across various tasks and models reveal clear distinctions in the causal distributions between normal behavior and misbehavior, enabling the development of accurate, lightweight detectors for a variety of misbehavior detection tasks.

♻ ☆ Diffusion-Reward Adversarial Imitation Learning

Chun-Mao Lai, Hsiang-Chun Wang, Ping-Chun Hsieh, Yu-Chiang Frank Wang, Min-Hung Chen, Shao-Hua Sun

Imitation learning aims to learn a policy from observing expert demonstrations without access to reward signals from environments. Generative adversarial imitation learning (GAIL) formulates imitation learning as adversarial learning, employing a generator policy learning to imitate expert behaviors and discriminator learning to distinguish the expert demonstrations from agent trajectories. Despite its encouraging results, GAIL training is often brittle and unstable. Inspired by the recent dominance of diffusion models in generative modeling, we propose Diffusion-Reward Adversarial Imitation Learning (DRAIL), which integrates a diffusion model into GAIL, aiming to yield more robust and smoother rewards for policy learning. Specifically, we propose a diffusion discriminative classifier to construct an enhanced discriminator, and design diffusion rewards based on the classifier's output for policy learning. Extensive experiments are conducted in navigation, manipulation, and locomotion, verifying DRAIL's effectiveness compared to prior imitation learning methods. Moreover, additional experimental results demonstrate the generalizability and data efficiency of DRAIL. Visualized learned reward functions of GAIL and DRAIL suggest that DRAIL can produce more robust and smoother rewards. Project page: https://nturobotlearninglab.github.io/DRAIL/

♻ ☆ Pruning By Explaining Revisited: Optimizing Attribution Methods to Prune CNNs and Transformers ECCV 2024

Sayed Mohammad Vakilzadeh Hatefi, Maximilian Dreyer, Reduan Achtibat, Thomas Wiegand, Wojciech Samek, Sebastian Lapuschkin

To solve ever more complex problems, Deep Neural Networks are scaled to billions of parameters, leading to huge computational costs. An effective approach to reduce computational requirements and increase efficiency is to prune unnecessary components of these often over-parameterized networks. Previous work has shown that attribution methods from the field of eXplainable AI serve as effective means to extract and prune the least relevant network components in a few-shot fashion. We extend the current state by proposing to explicitly optimize hyperparameters of attribution methods for the task of pruning, and further include transformer-based networks in our analysis. Our approach yields higher model compression rates of large transformer- and convolutional architectures (VGG, ResNet, ViT) compared to previous works, while still attaining high performance on ImageNet classification tasks. Here, our experiments indicate that transformers have a higher degree of over-parameterization compared to convolutional neural networks. Code is available at https://github.com/erfanhatefi/Pruning-by-eXplaining-in-PyTorch.

comment: Accepted as a workshop paper at ECCV 2024, 26 pages (11 pages manuscript, 3 pages references, 12 pages appendix)

♻ ☆ Correlated Proxies: A New Definition and Improved Mitigation for Reward Hacking

Cassidy Laidlaw, Shivam Singhal, Anca Dragan

Because it is difficult to precisely specify complex objectives, reinforcement learning policies are often optimized using flawed proxy rewards that seem to capture the true objective. However, optimizing proxy rewards frequently leads to reward hacking: the optimized reward function ceases to be a good proxy, and the resulting policy performs poorly with respect to the unspecified true reward. Principled solutions to reward hacking have been impeded by the lack of a good definition for the problem. To address this, we introduce a definition of reward hacking based on the correlation between proxy and true rewards for states and actions seen by a "base policy" that breaks down under optimization. We show that this definition captures reward hacking behavior across several realistic settings, including in reinforcement learning from human feedback (RLHF). We then show theoretically that regularization to the base policy can effectively prevent reward hacking. While current RLHF approaches apply a KL penalty between the action distributions of policies, our theory suggests that it is more effective to regularize using the $\chi^2$ divergence between the policies' occupancy measures. We intuitively show why this type of regularization is superior and demonstrate that it better mitigates reward hacking in practice across four realistic domains, including RLHF for LLMs. Our code is available at https://github.com/cassidylaidlaw/orpo.

♻ ☆ Conditional Language Policy: A General Framework for Steerable Multi-Objective Finetuning EMNLP 2024

Kaiwen Wang, Rahul Kidambi, Ryan Sullivan, Alekh Agarwal, Christoph Dann, Andrea Michi, Marco Gelmi, Yunxuan Li, Raghav Gupta, Avinava Dubey, Alexandre Ramé, Johan Ferret, Geoffrey Cideron, Le Hou, Hongkun Yu, Amr Ahmed, Aranyak Mehta, Léonard Hussenot, Olivier Bachem, Edouard Leurent

Reward-based finetuning is crucial for aligning language policies with intended behaviors (e.g., creativity and safety). A key challenge is to develop steerable language models that trade-off multiple (conflicting) objectives in a flexible and efficient manner. This paper presents Conditional Language Policy (CLP), a general framework for finetuning language models on multiple objectives. Building on techniques from multi-task training and parameter-efficient finetuning, CLP learn steerable models that effectively trade-off conflicting objectives at inference time. Notably, this does not require training or maintaining multiple models to achieve different trade-offs between the objectives. Through extensive experiments and ablations on two summarization datasets, we show that CLP learns steerable language models that outperform and Pareto-dominate the existing approaches for multi-objective finetuning.

comment: 40 pages. Findings of EMNLP 2024

♻ ☆ VILA-U: a Unified Foundation Model Integrating Visual Understanding and Generation

Yecheng Wu, Zhuoyang Zhang, Junyu Chen, Haotian Tang, Dacheng Li, Yunhao Fang, Ligeng Zhu, Enze Xie, Hongxu Yin, Li Yi, Song Han, Yao Lu

VILA-U is a Unified foundation model that integrates Video, Image, Language understanding and generation. Traditional visual language models (VLMs) use separate modules for understanding and generating visual content, which can lead to misalignment and increased complexity. In contrast, VILA-U employs a single autoregressive next-token prediction framework for both tasks, eliminating the need for additional components like diffusion models. This approach not only simplifies the model but also achieves near state-of-the-art performance in visual language understanding and generation. The success of VILA-U is attributed to two main factors: the unified vision tower that aligns discrete visual tokens with textual inputs during pretraining, which enhances visual perception, and autoregressive image generation can achieve similar quality as diffusion models with high-quality dataset. This allows VILA-U to perform comparably to more complex models using a fully token-based autoregressive framework.

comment: Code: https://github.com/mit-han-lab/vila-u. The first two authors contributed equally to this work

♻ ☆ Counter-Current Learning: A Biologically Plausible Dual Network Approach for Deep Learning NeurIPS 2024

Chia-Hsiang Kao, Bharath Hariharan

Despite its widespread use in neural networks, error backpropagation has faced criticism for its lack of biological plausibility, suffering from issues such as the backward locking problem and the weight transport problem. These limitations have motivated researchers to explore more biologically plausible learning algorithms that could potentially shed light on how biological neural systems adapt and learn. Inspired by the counter-current exchange mechanisms observed in biological systems, we propose counter-current learning (CCL), a biologically plausible framework for credit assignment in neural networks. This framework employs a feedforward network to process input data and a feedback network to process targets, with each network enhancing the other through anti-parallel signal propagation. By leveraging the more informative signals from the bottom layer of the feedback network to guide the updates of the top layer of the feedforward network and vice versa, CCL enables the simultaneous transformation of source inputs to target outputs and the dynamic mutual influence of these transformations. Experimental results on MNIST, FashionMNIST, CIFAR10, and CIFAR100 datasets using multi-layer perceptrons and convolutional neural networks demonstrate that CCL achieves comparable performance to other biologically plausible algorithms while offering a more biologically realistic learning mechanism. Furthermore, we showcase the applicability of our approach to an autoencoder task, underscoring its potential for unsupervised representation learning. Our work presents a direction for biologically inspired and plausible learning algorithms, offering an alternative mechanism of learning and adaptation in neural networks.

comment: Accepted at NeurIPS 2024. Code available at https://github.com/IandRover/CCL-NeurIPS24

♻ ☆ Proof of Thought : Neurosymbolic Program Synthesis allows Robust and Interpretable Reasoning NeurIPS 2024

Debargha Ganguly, Srinivasan Iyengar, Vipin Chaudhary, Shivkumar Kalyanaraman

Large Language Models (LLMs) have revolutionized natural language processing, yet they struggle with inconsistent reasoning, particularly in novel domains and complex logical sequences. This research introduces Proof of Thought, a framework that enhances the reliability and transparency of LLM outputs. Our approach bridges LLM-generated ideas with formal logic verification, employing a custom interpreter to convert LLM outputs into First Order Logic constructs for theorem prover scrutiny. Central to our method is an intermediary JSON-based Domain-Specific Language, which by design balances precise logical structures with intuitive human concepts. This hybrid representation enables both rigorous validation and accessible human comprehension of LLM reasoning processes. Key contributions include a robust type system with sort management for enhanced logical integrity, explicit representation of rules for clear distinction between factual and inferential knowledge, and a flexible architecture that allows for easy extension to various domain-specific applications. We demonstrate Proof of Thought's effectiveness through benchmarking on StrategyQA and a novel multimodal reasoning task, showing improved performance in open-ended scenarios. By providing verifiable and interpretable results, our technique addresses critical needs for AI system accountability and sets a foundation for human-in-the-loop oversight in high-stakes domains.

comment: 38th Conference on Neural Information Processing Systems (NeurIPS 2024) System 2 Reasoning At Scale Workshop

♻ ☆ AlleNoise: large-scale text classification benchmark dataset with real-world label noise

Alicja Rączkowska, Aleksandra Osowska-Kurczab, Jacek Szczerbiński, Kalina Jasinska-Kobus, Klaudia Nazarko

Label noise remains a challenge for training robust classification models. Most methods for mitigating label noise have been benchmarked using primarily datasets with synthetic noise. While the need for datasets with realistic noise distribution has partially been addressed by web-scraped benchmarks such as WebVision and Clothing1M, those benchmarks are restricted to the computer vision domain. With the growing importance of Transformer-based models, it is crucial to establish text classification benchmarks for learning with noisy labels. In this paper, we present AlleNoise, a new curated text classification benchmark dataset with real-world instance-dependent label noise, containing over 500,000 examples across approximately 5,600 classes, complemented with a meaningful, hierarchical taxonomy of categories. The noise distribution comes from actual users of a major e-commerce marketplace, so it realistically reflects the semantics of human mistakes. In addition to the noisy labels, we provide human-verified clean labels, which help to get a deeper insight into the noise distribution, unlike web-scraped datasets typically used in the field. We demonstrate that a representative selection of established methods for learning with noisy labels is inadequate to handle such real-world noise. In addition, we show evidence that these algorithms do not alleviate excessive memorization. As such, with AlleNoise, we set the bar high for the development of label noise methods that can handle real-world label noise in text classification tasks. The code and dataset are available for download at https://github.com/allegro/AlleNoise.

♻ ☆ CondTSF: One-line Plugin of Dataset Condensation for Time Series Forecasting NeurIPS 2024

Jianrong Ding, Zhanyu Liu, Guanjie Zheng, Haiming Jin, Linghe Kong

Dataset condensation is a newborn technique that generates a small dataset that can be used in training deep neural networks to lower training costs. The objective of dataset condensation is to ensure that the model trained with the synthetic dataset can perform comparably to the model trained with full datasets. However, existing methods predominantly concentrate on classification tasks, posing challenges in their adaptation to time series forecasting (TS-forecasting). This challenge arises from disparities in the evaluation of synthetic data. In classification, the synthetic data is considered well-distilled if the model trained with the full dataset and the model trained with the synthetic dataset yield identical labels for the same input, regardless of variations in output logits distribution. Conversely, in TS-forecasting, the effectiveness of synthetic data distillation is determined by the distance between predictions of the two models. The synthetic data is deemed well-distilled only when all data points within the predictions are similar. Consequently, TS-forecasting has a more rigorous evaluation methodology compared to classification. To mitigate this gap, we theoretically analyze the optimization objective of dataset condensation for TS-forecasting and propose a new one-line plugin of dataset condensation designated as Dataset Condensation for Time Series Forecasting (CondTSF) based on our analysis. Plugging CondTSF into previous dataset condensation methods facilitates a reduction in the distance between the predictions of the model trained with the full dataset and the model trained with the synthetic dataset, thereby enhancing performance. We conduct extensive experiments on eight commonly used time series datasets. CondTSF consistently improves the performance of all previous dataset condensation methods across all datasets, particularly at low condensing ratios.

comment: Accepted by NeurIPS 2024, the project can be found at https://github.com/RafaDD/CondTSF

♻ ☆ Learning a quantum computer's capability

Daniel Hothem, Kevin Young, Tommie Catanach, Timothy Proctor

Accurately predicting a quantum computer's capability -- which circuits it can run and how well it can run them -- is a foundational goal of quantum characterization and benchmarking. As modern quantum computers become increasingly hard to simulate, we must develop accurate and scalable predictive capability models to help researchers and stakeholders decide which quantum computers to build and use. In this work, we propose a hardware-agnostic method to efficiently construct scalable predictive models of a quantum computer's capability for almost any class of circuits, and demonstrate our method using convolutional neural networks (CNNs). Our CNN-based approach works by efficiently representing a circuit as a three-dimensional tensor and then using a CNN to predict its success rate. Our CNN capability models obtain approximately a $1\%$ average absolute prediction error when modeling processors experiencing both Markovian and non-Markovian stochastic Pauli errors. We also apply our CNNs to model the capabilities of cloud-access quantum computing systems, obtaining moderate prediction accuracy (average absolute error around $2-5\%$), and we highlight the challenges to building better neural network capability models.

comment: 20 pages, 11 figures, plus appendices

♻ ☆ Federated Class-Incremental Learning with Hierarchical Generative Prototypes

Riccardo Salami, Pietro Buzzega, Matteo Mosconi, Mattia Verasani, Simone Calderara

Federated Learning (FL) aims at unburdening the training of deep models by distributing computation across multiple devices (clients) while safeguarding data privacy. On top of that, Federated Continual Learning (FCL) also accounts for data distribution evolving over time, mirroring the dynamic nature of real-world environments. While previous studies have identified Catastrophic Forgetting and Client Drift as primary causes of performance degradation in FCL, we shed light on the importance of Incremental Bias and Federated Bias, which cause models to prioritize classes that are recently introduced or locally predominant, respectively. Our proposal constrains both biases in the last layer by efficiently finetuning a pre-trained backbone using learnable prompts, resulting in clients that produce less biased representations and more biased classifiers. Therefore, instead of solely relying on parameter aggregation, we leverage generative prototypes to effectively balance the predictions of the global model. Our method significantly improves the current State Of The Art, providing an average increase of +7.8% in accuracy. Code to reproduce the results is provided in the suppl. material.

♻ ☆ Do Large Language Models Truly Grasp Mathematics? An Empirical Exploration

Wei Xie, Shuoyoucheng Ma, Zhenhua Wang, Enze Wang, Baosheng Wang, Jinshu Su

Despite their proficiency in math tasks, the mechanisms underlying LLMs' mathematical reasoning abilities remain a subject of debate. Recent studies suggest that chain-of-thought (CoT) prompts can bolster mathematical reasoning by encouraging LLMs to employ human-like logical reasoning (System 2), enabling them to excel on the Cognitive Reflection Test (CRT). To assess whether LLMs genuinely possess System 2-like logical reasoning, we introduced targeted modifications to CRT problems. Our findings reveal that, despite the use of CoT prompts, mainstream LLMs, including the latest o1-preview model, continue to exhibit a significant error rate. Further analysis indicates that they predominantly rely on System 1-like intuitive reasoning and pattern matching derived from training data, rather than demonstrating mastery of mathematical thinking. This discovery challenges the prevailing notion that LLMs possess genuine logical reasoning abilities and that CoT can enhance them. Consequently, this work may temper overly optimistic projections regarding LLMs' advancement toward artificial general intelligence.

♻ ☆ On the potential of Optimal Transport in Geospatial Data Science

Nina Wiedemann, Théo Uscidda, Martin Raubal

Prediction problems in geographic information science and transportation are often motivated by the possibility to enhance operational efficiency and thereby reduce emissions. Examples range from predicting car sharing demand for relocation planning to forecasting traffic congestion for navigation purposes. However, conventional accuracy metrics ignore the spatial distribution of the errors, despite its relevance for operations. Here, we put forward a spatially aware evaluation metric and loss function based on Optimal Transport (OT). Our framework leverages partial OT and can minimize relocation costs in any spatial prediction problem. We showcase the advantages of OT-based evaluation over conventional metrics and further demonstrate the application of an OT loss function for improving forecasts of bike sharing demand and charging station occupancy. Thus, our framework not only aligns with operational considerations, but also signifies a step forward in refining predictions within geospatial applications. All code is available at https://github.com/mie-lab/geospatialOT.

♻ ☆ Quantum Architecture Search with Unsupervised Representation Learning

Yize Sun, Zixin Wu, Yunpu Ma, Volker Tresp

Unsupervised representation learning presents new opportunities for advancing Quantum Architecture Search (QAS) on Noisy Intermediate-Scale Quantum (NISQ) devices. QAS is designed to optimize quantum circuits for Variational Quantum Algorithms (VQAs). Most QAS algorithms tightly couple the search space and search algorithm, typically requiring the evaluation of numerous quantum circuits, resulting in high computational costs and limiting scalability to larger quantum circuits. Predictor-based QAS algorithms mitigate this issue by estimating circuit performance based on structure or embedding. However, these methods often demand time-intensive labeling to optimize gate parameters across many circuits, which is crucial for training accurate predictors. Inspired by the classical neural architecture search algorithm Arch2vec, we investigate the potential of unsupervised representation learning for QAS without relying on predictors. Our framework decouples unsupervised architecture representation learning from the search process, enabling the learned representations to be applied across various downstream tasks. Additionally, it integrates an improved quantum circuit graph encoding scheme, addressing the limitations of existing representations and enhancing search efficiency. This predictor-free approach removes the need for large labeled datasets. During the search, we employ REINFORCE and Bayesian Optimization to explore the latent representation space and compare their performance against baseline methods. Our results demonstrate that the framework efficiently identifies high-performing quantum circuits with fewer search iterations.

comment: 9 Pages, quantum architecture search, unsupervised representation learning

♻ ☆ MOTIVE: A Drug-Target Interaction Graph For Inductive Link Prediction

John Arevalo, Ellen Su, Anne E Carpenter, Shantanu Singh

Drug-target interaction (DTI) prediction is crucial for identifying new therapeutics and detecting mechanisms of action. While structure-based methods accurately model physical interactions between a drug and its protein target, cell-based assays such as Cell Painting can better capture complex DTI interactions. This paper introduces MOTIVE, a Morphological cOmpound Target Interaction Graph dataset comprising Cell Painting features for 11,000 genes and 3,600 compounds, along with their relationships extracted from seven publicly available databases. We provide random, cold-source (new drugs), and cold-target (new genes) data splits to enable rigorous evaluation under realistic use cases. Our benchmark results show that graph neural networks that use Cell Painting features consistently outperform those that learn from graph structure alone, feature-based models, and topological heuristics. MOTIVE accelerates both graph ML research and drug discovery by promoting the development of more reliable DTI prediction models. MOTIVE resources are available at https://github.com/carpenter-singh-lab/motive.

♻ ☆ Bounded KRnet and its applications to density estimation and approximation

Li Zeng, Xiaoliang Wan, Tao Zhou

In this paper, we develop an invertible mapping, called B-KRnet, on a bounded domain and apply it to density estimation/approximation for data or the solutions of PDEs such as the Fokker-Planck equation and the Keller-Segel equation. Similar to KRnet, the structure of B-KRnet adapts the pseudo-triangular structure into a normalizing flow model. The main difference between B-KRnet and KRnet is that B-KRnet is defined on a hypercube while KRnet is defined on the whole space, in other words, a new mechanism is introduced in B-KRnet to maintain the exact invertibility. Using B-KRnet as a transport map, we obtain an explicit probability density function (PDF) model that corresponds to the pushforward of a prior (uniform) distribution on the hypercube. It can be directly applied to density estimation when only data are available. By coupling KRnet and B-KRnet, we define a deep generative model on a high-dimensional domain where some dimensions are bounded and other dimensions are unbounded. A typical case is the solution of the stationary kinetic Fokker-Planck equation, which is a PDF of position and momentum. Based on B-KRnet, we develop an adaptive learning approach to approximate partial differential equations whose solutions are PDFs or can be treated as PDFs. A variety of numerical experiments is presented to demonstrate the effectiveness of B-KRnet.

comment: 26 pages, 13 figures

♻ ☆ Linear Adversarial Concept Erasure ICML 2022

Shauli Ravfogel, Michael Twiton, Yoav Goldberg, Ryan Cotterell

Modern neural models trained on textual data rely on pre-trained representations that emerge without direct supervision. As these representations are increasingly being used in real-world applications, the inability to \emph{control} their content becomes an increasingly important problem. We formulate the problem of identifying and erasing a linear subspace that corresponds to a given concept, in order to prevent linear predictors from recovering the concept. We model this problem as a constrained, linear maximin game, and show that existing solutions are generally not optimal for this task. We derive a closed-form solution for certain objectives, and propose a convex relaxation, \method, that works well for others. When evaluated in the context of binary gender removal, the method recovers a low-dimensional subspace whose removal mitigates bias by intrinsic and extrinsic evaluation. We show that the method is highly expressive, effectively mitigating bias in deep nonlinear classifiers while maintaining tractability and interpretability.

comment: Accepted in ICML 2022; a revised version

♻ ☆ Accessible, At-Home Detection of Parkinson's Disease via Multi-task Video Analysis

Md Saiful Islam, Tariq Adnan, Jan Freyberg, Sangwu Lee, Abdelrahman Abdelkader, Meghan Pawlik, Cathe Schwartz, Karen Jaffe, Ruth B. Schneider, E Ray Dorsey, Ehsan Hoque

Limited accessibility to neurological care leads to underdiagnosed Parkinson's Disease (PD), preventing early intervention. Existing AI-based PD detection methods primarily focus on unimodal analysis of motor or speech tasks, overlooking the multifaceted nature of the disease. To address this, we introduce a large-scale, multi-task video dataset consisting of 1102 sessions (each containing videos of finger tapping, facial expression, and speech tasks captured via webcam) from 845 participants (272 with PD). We propose a novel Uncertainty-calibrated Fusion Network (UFNet) that leverages this multimodal data to enhance diagnostic accuracy. UFNet employs independent task-specific networks, trained with Monte Carlo Dropout for uncertainty quantification, followed by self-attended fusion of features, with attention weights dynamically adjusted based on task-specific uncertainties. To ensure patient-centered evaluation, the participants were randomly split into three sets: 60% for training, 20% for model selection, and 20% for final performance evaluation. UFNet significantly outperformed single-task models in terms of accuracy, area under the ROC curve (AUROC), and sensitivity while maintaining non-inferior specificity. Withholding uncertain predictions further boosted the performance, achieving 88.0+-0.3%$ accuracy, 93.0+-0.2% AUROC, 79.3+-0.9% sensitivity, and 92.6+-0.3% specificity, at the expense of not being able to predict for 2.3+-0.3% data (+- denotes 95% confidence interval). Further analysis suggests that the trained model does not exhibit any detectable bias across sex and ethnic subgroups and is most effective for individuals aged between 50 and 80. Requiring only a webcam and microphone, our approach facilitates accessible home-based PD screening, especially in regions with limited healthcare resources.

♻ ☆ Certifiably Robust Policies for Uncertain Parametric Environments

Yannik Schnitzer, Alessandro Abate, David Parker

We present a data-driven approach for producing policies that are provably robust across unknown stochastic environments. Existing approaches can learn models of a single environment as an interval Markov decision processes (IMDP) and produce a robust policy with a probably approximately correct (PAC) guarantee on its performance. However these are unable to reason about the impact of environmental parameters underlying the uncertainty. We propose a framework based on parametric Markov decision processes (MDPs) with unknown distributions over parameters. We learn and analyse IMDPs for a set of unknown sample environments induced by parameters. The key challenge is then to produce meaningful performance guarantees that combine the two layers of uncertainty: (1) multiple environments induced by parameters with an unknown distribution; (2) unknown induced environments which are approximated by IMDPs. We present a novel approach based on scenario optimisation that yields a single PAC guarantee quantifying the risk level for which a specified performance level can be assured in unseen environments, plus a means to trade-off risk and performance. We implement and evaluate our framework using multiple robust policy generation methods on a range of benchmarks. We show that our approach produces tight bounds on a policy's performance with high confidence.

♻ ☆ On provable privacy vulnerabilities of graph representations

Ruofan Wu, Guanhua Fang, Qiying Pan, Mingyang Zhang, Tengfei Liu, Weiqiang Wang

Graph representation learning (GRL) is critical for extracting insights from complex network structures, but it also raises security concerns due to potential privacy vulnerabilities in these representations. This paper investigates the structural vulnerabilities in graph neural models where sensitive topological information can be inferred through edge reconstruction attacks. Our research primarily addresses the theoretical underpinnings of similarity-based edge reconstruction attacks (SERA), furnishing a non-asymptotic analysis of their reconstruction capacities. Moreover, we present empirical corroboration indicating that such attacks can perfectly reconstruct sparse graphs as graph size increases. Conversely, we establish that sparsity is a critical factor for SERA's effectiveness, as demonstrated through analysis and experiments on (dense) stochastic block models. Finally, we explore the resilience of private graph representations produced via noisy aggregation (NAG) mechanism against SERA. Through theoretical analysis and empirical assessments, we affirm the mitigation of SERA using NAG . In parallel, we also empirically delineate instances wherein SERA demonstrates both efficacy and deficiency in its capacity to function as an instrument for elucidating the trade-off between privacy and utility.

♻ ☆ Adaptive Variance Reduction for Stochastic Optimization under Weaker Assumptions

Wei Jiang, Sifan Yang, Yibo Wang, Lijun Zhang

This paper explores adaptive variance reduction methods for stochastic optimization based on the STORM technique. Existing adaptive extensions of STORM rely on strong assumptions like bounded gradients and bounded function values, or suffer an additional $\mathcal{O}(\log T)$ term in the convergence rate. To address these limitations, we introduce a novel adaptive STORM method that achieves an optimal convergence rate of $\mathcal{O}(T^{-1/3})$ for non-convex functions with our newly designed learning rate strategy. Compared with existing approaches, our method requires weaker assumptions and attains the optimal convergence rate without the additional $\mathcal{O}(\log T)$ term. We also extend the proposed technique to stochastic compositional optimization, obtaining the same optimal rate of $\mathcal{O}(T^{-1/3})$. Furthermore, we investigate the non-convex finite-sum problem and develop another innovative adaptive variance reduction method that achieves an optimal convergence rate of $\mathcal{O}(n^{1/4} T^{-1/2} )$, where $n$ represents the number of component functions. Numerical experiments across various tasks validate the effectiveness of our method.

comment: arXiv admin note: substantial text overlap with arXiv:2406.00489

♻ ☆ Posterior Sampling-based Online Learning for Episodic POMDPs

Dengwang Tang, Dongze Ye, Rahul Jain, Ashutosh Nayyar, Pierluigi Nuzzo

Learning in POMDPs is known to be significantly harder than in MDPs. In this paper, we consider the online learning problem for episodic POMDPs with unknown transition and observation models. We propose a Posterior Sampling-based reinforcement learning algorithm for POMDPs (PS4POMDPs), which is much simpler and more implementable compared to state-of-the-art optimism-based online learning algorithms for POMDPs. We show that the Bayesian regret of the proposed algorithm scales as the square root of the number of episodes and is polynomial in the other parameters. In a general setting, the regret scales exponentially in the horizon length $H$, and we show that this is inevitable by providing a lower bound. However, when the POMDP is undercomplete and weakly revealing (a common assumption in the recent literature), we establish a polynomial Bayesian regret bound. We finally propose a posterior sampling algorithm for multi-agent POMDPs, and show it too has sublinear regret.

comment: 41 pages, 9 figures

♻ ☆ Efficient Sign-Based Optimization: Accelerating Convergence via Variance Reduction

Wei Jiang, Sifan Yang, Wenhao Yang, Lijun Zhang

Sign stochastic gradient descent (signSGD) is a communication-efficient method that transmits only the sign of stochastic gradients for parameter updating. Existing literature has demonstrated that signSGD can achieve a convergence rate of $\mathcal{O}(d^{1/2}T^{-1/4})$, where $d$ represents the dimension and $T$ is the iteration number. In this paper, we improve this convergence rate to $\mathcal{O}(d^{1/2}T^{-1/3})$ by introducing the Sign-based Stochastic Variance Reduction (SSVR) method, which employs variance reduction estimators to track gradients and leverages their signs to update. For finite-sum problems, our method can be further enhanced to achieve a convergence rate of $\mathcal{O}(m^{1/4}d^{1/2}T^{-1/2})$, where $m$ denotes the number of component functions. Furthermore, we investigate the heterogeneous majority vote in distributed settings and introduce two novel algorithms that attain improved convergence rates of $\mathcal{O}(d^{1/2}T^{-1/2} + dn^{-1/2})$ and $\mathcal{O}(d^{1/4}T^{-1/4})$ respectively, outperforming the previous results of $\mathcal{O}(dT^{-1/4} + dn^{-1/2})$ and $\mathcal{O}(d^{3/8}T^{-1/8})$, where $n$ represents the number of nodes. Numerical experiments across different tasks validate the effectiveness of our proposed methods.

♻ ☆ Anomaly Prediction: A Novel Approach with Explicit Delay and Horizon

Jiang You, Arben Cela, René Natowicz, Jacob Ouanounou, Patrick Siarry

Anomaly detection in time series data is a critical challenge across various domains. Traditional methods typically focus on identifying anomalies in immediate subsequent steps, often underestimating the significance of temporal dynamics such as delay time and horizons of anomalies, which generally require extensive post-analysis. This paper introduces a novel approach for time series anomaly prediction, incorporating temporal information directly into the prediction results. We propose a new dataset specifically designed to evaluate this approach and conduct comprehensive experiments using several state-of-the-art methods. Our results demonstrate the efficacy of our approach in providing timely and accurate anomaly predictions, setting a new benchmark for future research in this field.

♻ ☆ Acquiring Better Load Estimates by Combining Anomaly and Change Point Detection in Power Grid Time-series Measurements

Roel Bouman, Linda Schmeitz, Luco Buise, Jacco Heres, Yuliya Shapovalova, Tom Heskes

In this paper we present novel methodology for automatic anomaly and switch event filtering to improve load estimation in power grid systems. By leveraging unsupervised methods with supervised optimization, our approach prioritizes interpretability while ensuring robust and generalizable performance on unseen data. Through experimentation, a combination of binary segmentation for change point detection and statistical process control for anomaly detection emerges as the most effective strategy, specifically when ensembled in a novel sequential manner. Results indicate the clear wasted potential when filtering is not applied. The automatic load estimation is also fairly accurate, with approximately 90% of estimates falling within a 10% error margin, with only a single significant failure in both the minimum and maximum load estimates across 60 measurements in the test set. Our methodology's interpretability makes it particularly suitable for critical infrastructure planning, thereby enhancing decision-making processes.

comment: All code can be found at: https://github.com/RoelBouman/StormPhase2

♻ ☆ A spring-block theory of feature learning in deep neural networks

Cheng Shi, Liming Pan, Ivan Dokmanić

Feature-learning deep nets progressively collapse data to a regular low-dimensional geometry. How this phenomenon emerges from collective action of nonlinearity, noise, learning rate, and other choices that shape the dynamics, has eluded first-principles theories built from microscopic neuronal dynamics. We exhibit a noise-nonlinearity phase diagram that identifies regimes where shallow or deep layers learn more effectively. We then propose a macroscopic mechanical theory that reproduces the diagram, explaining why some DNNs are lazy and some active, and linking feature learning across layers to generalization.

♻ ☆ CPE-Pro: A Structure-Sensitive Deep Learning Method for Protein Representation and Origin Evaluation

Wenrui Gou, Wenhui Ge, Yang Tan, Mingchen Li, Guisheng Fan, Huiqun Yu

Protein structures are important for understanding their functions and interactions. Currently, many protein structure prediction methods are enriching the structure database. Discriminating the origin of structures is crucial for distinguishing between experimentally resolved and computationally predicted structures, evaluating the reliability of prediction methods, and guiding downstream biological studies. Building on works in structure prediction, We developed a structure-sensitive supervised deep learning model, Crystal vs Predicted Evaluator for Protein Structure (CPE-Pro), to represent and discriminate the origin of protein structures. CPE-Pro learns the structural information of proteins and captures inter-structural differences to achieve accurate traceability on four data classes, and is expected to be extended to more. Simultaneously, we utilized Foldseek to encode protein structures into "structure-sequences" and trained a protein Structural Sequence Language Model, SSLM. Preliminary experiments demonstrated that, compared to large-scale protein language models pre-trained on vast amounts of amino acid sequences, the "structure-sequence" enables the language model to learn more informative protein features, enhancing and optimizing structural representations. We have provided the code, model weights, and all related materials on https://github.com/GouWenrui/CPE-Pro-main.git.

♻ ☆ Towards Croppable Implicit Neural Representations NeurIPS 2024

Maor Ashkenazi, Eran Treister

Implicit Neural Representations (INRs) have peaked interest in recent years due to their ability to encode natural signals using neural networks. While INRs allow for useful applications such as interpolating new coordinates and signal compression, their black-box nature makes it difficult to modify them post-training. In this paper we explore the idea of editable INRs, and specifically focus on the widely used cropping operation. To this end, we present Local-Global SIRENs -- a novel INR architecture that supports cropping by design. Local-Global SIRENs are based on combining local and global feature extraction for signal encoding. What makes their design unique is the ability to effortlessly remove specific portions of an encoded signal, with a proportional weight decrease. This is achieved by eliminating the corresponding weights from the network, without the need for retraining. We further show how this architecture can be used to support the straightforward extension of previously encoded signals. Beyond signal editing, we examine how the Local-Global approach can accelerate training, enhance encoding of various signals, improve downstream performance, and be applied to modern INRs such as INCODE, highlighting its potential and flexibility. Code is available at https://github.com/maorash/Local-Global-INRs.

comment: Accepted to NeurIPS 2024

♻ ☆ Breaking Class Barriers: Efficient Dataset Distillation via Inter-Class Feature Compensator

Xin Zhang, Jiawei Du, Ping Liu, Joey Tianyi Zhou

Dataset distillation has emerged as a technique aiming to condense informative features from large, natural datasets into a compact and synthetic form. While recent advancements have refined this technique, its performance is bottlenecked by the prevailing class-specific synthesis paradigm. Under this paradigm, synthetic data is optimized exclusively for a pre-assigned one-hot label, creating an implicit class barrier in feature condensation. This leads to inefficient utilization of the distillation budget and oversight of inter-class feature distributions, which ultimately limits the effectiveness and efficiency, as demonstrated in our analysis. To overcome these constraints, this paper presents the Inter-class Feature Compensator (INFER), an innovative distillation approach that transcends the class-specific data-label framework widely utilized in current dataset distillation methods. Specifically, INFER leverages a Universal Feature Compensator (UFC) to enhance feature integration across classes, enabling the generation of multiple additional synthetic instances from a single UFC input. This significantly improves the efficiency of the distillation budget. Moreover, INFER enriches inter-class interactions during the distillation, thereby enhancing the effectiveness and generalizability of the distilled data. By allowing for the linear interpolation of labels similar to those in the original dataset, INFER meticulously optimizes the synthetic data and dramatically reduces the size of soft labels in the synthetic dataset to almost zero, establishing a new benchmark for efficiency and effectiveness in dataset distillation.

♻ ☆ I've Got 99 Problems But FLOPS Ain't One

Alexandru M. Gherghescu, Vlad-Andrei Bădoiu, Alexandru Agache, Mihai-Valentin Dumitru, Iuliu Vasilescu, Radu Mantu, Costin Raiciu

Hyperscalers dominate the landscape of large network deployments, yet they rarely share data or insights about the challenges they face. In light of this supremacy, what problems can we find to solve in this space? We take an unconventional approach to find relevant research directions, starting from public plans to build a $100 billion datacenter for machine learning applications. Leveraging the language models scaling laws, we discover what workloads such a datacenter might carry and explore the challenges one may encounter in doing so, with a focus on networking research. We conclude that building the datacenter and training such models is technically possible, but this requires novel wide-area transports for inter-DC communication, a multipath transport and novel datacenter topologies for intra-datacenter communication, high speed scale-up networks and transports, outlining a rich research agenda for the networking community.

♻ ☆ TargetCall: Eliminating the Wasted Computation in Basecalling via Pre-Basecalling Filtering

Meryem Banu Cavlak, Gagandeep Singh, Mohammed Alser, Can Firtina, Joël Lindegger, Mohammad Sadrosadati, Nika Mansouri Ghiasi, Can Alkan, Onur Mutlu

Basecalling is an essential step in nanopore sequencing analysis where the raw signals of nanopore sequencers are converted into nucleotide sequences, i.e., reads. State-of-the-art basecallers employ complex deep learning models to achieve high basecalling accuracy. This makes basecalling computationally inefficient and memory-hungry, bottlenecking the entire genome analysis pipeline. However, for many applications, the majority of reads do no match the reference genome of interest (i.e., target reference) and thus are discarded in later steps in the genomics pipeline, wasting the basecalling computation. To overcome this issue, we propose TargetCall, the first pre-basecalling filter to eliminate the wasted computation in basecalling. TargetCall's key idea is to discard reads that will not match the target reference (i.e., off-target reads) prior to basecalling. TargetCall consists of two main components: (1) LightCall, a lightweight neural network basecaller that produces noisy reads; and (2) Similarity Check, which labels each of these noisy reads as on-target or off-target by matching them to the target reference. Our thorough experimental evaluations show that TargetCall 1) improves the end-to-end basecalling runtime performance of the state-of-the-art basecaller by 3.31x while maintaining high (98.88%) recall in keeping on-target reads, 2) maintains high accuracy in downstream analysis, and 3) achieves better runtime performance, throughput, recall, precision, and generality compared to prior works. TargetCall is available at https://github.com/CMU-SAFARI/TargetCall.

♻ ☆ Interpreting Context Look-ups in Transformers: Investigating Attention-MLP Interactions EMNLP 2024

Clement Neo, Shay B. Cohen, Fazl Barez

Understanding the inner workings of large language models (LLMs) is crucial for advancing their theoretical foundations and real-world applications. While the attention mechanism and multi-layer perceptrons (MLPs) have been studied independently, their interactions remain largely unexplored. This study investigates how attention heads and next-token neurons interact in LLMs to predict new words. We propose a methodology to identify next-token neurons, find prompts that highly activate them, and determine the upstream attention heads responsible. We then generate and evaluate explanations for the activity of these attention heads in an automated manner. Our findings reveal that some attention heads recognize specific contexts relevant to predicting a token and activate a downstream token-predicting neuron accordingly. This mechanism provides a deeper understanding of how attention heads work with MLP neurons to perform next-token prediction. Our approach offers a foundation for further research into the intricate workings of LLMs and their impact on text generation and understanding.

comment: Accepted to EMNLP 2024 Main Conference

♻ ☆ Ornstein-Uhlenbeck Adaptation as a Mechanism for Learning in Brains and Machines

Jesus Garcia Fernandez, Nasir Ahmad, Marcel van Gerven

Learning is a fundamental property of intelligent systems, observed across biological organisms and engineered systems. While modern intelligent systems typically rely on gradient descent for learning, the need for exact gradients and complex information flow makes its implementation in biological and neuromorphic systems challenging. This has motivated the exploration of alternative learning mechanisms that can operate locally and do not rely on exact gradients. In this work, we introduce a novel approach that leverages noise in the parameters of the system and global reinforcement signals. Using an Ornstein-Uhlenbeck process with adaptive dynamics, our method balances exploration and exploitation during learning, driven by deviations from error predictions, akin to reward prediction error. Operating in continuous time, Orstein-Uhlenbeck adaptation (OUA) is proposed as a general mechanism for learning dynamic, time-evolving environments. We validate our approach across diverse tasks, including supervised learning and reinforcement learning in feedforward and recurrent systems. Additionally, we demonstrate that it can perform meta-learning, adjusting hyper-parameters autonomously. Our results indicate that OUA provides a viable alternative to traditional gradient-based methods, with potential applications in neuromorphic computing. It also hints at a possible mechanism for noise-driven learning in the brain, where stochastic neurotransmitter release may guide synaptic adjustments.

♻ ☆ FOOGD: Federated Collaboration for Both Out-of-distribution Generalization and Detection NeurIPS 2024

Xinting Liao, Weiming Liu, Pengyang Zhou, Fengyuan Yu, Jiahe Xu, Jun Wang, Wenjie Wang, Chaochao Chen, Xiaolin Zheng

Federated learning (FL) is a promising machine learning paradigm that collaborates with client models to capture global knowledge. However, deploying FL models in real-world scenarios remains unreliable due to the coexistence of in-distribution data and unexpected out-of-distribution (OOD) data, such as covariate-shift and semantic-shift data. Current FL researches typically address either covariate-shift data through OOD generalization or semantic-shift data via OOD detection, overlooking the simultaneous occurrence of various OOD shifts. In this work, we propose FOOGD, a method that estimates the probability density of each client and obtains reliable global distribution as guidance for the subsequent FL process. Firstly, SM3D in FOOGD estimates score model for arbitrary distributions without prior constraints, and detects semantic-shift data powerfully. Then SAG in FOOGD provides invariant yet diverse knowledge for both local covariate-shift generalization and client performance generalization. In empirical validations, FOOGD significantly enjoys three main advantages: (1) reliably estimating non-normalized decentralized distributions, (2) detecting semantic shift data via score values, and (3) generalizing to covariate-shift data by regularizing feature extractor. The prejoct is open in https://github.com/XeniaLLL/FOOGD-main.git.

comment: NeurIPS 2024

♻ ☆ Few-Shot Adversarial Prompt Learning on Vision-Language Models NeurIPS 2024

Yiwei Zhou, Xiaobo Xia, Zhiwei Lin, Bo Han, Tongliang Liu

The vulnerability of deep neural networks to imperceptible adversarial perturbations has attracted widespread attention. Inspired by the success of vision-language foundation models, previous efforts achieved zero-shot adversarial robustness by aligning adversarial visual features with text supervision. However, in practice, they are still unsatisfactory due to several issues, including heavy adaptation cost, suboptimal text supervision, and uncontrolled natural generalization capacity. In this paper, to address these issues, we propose a few-shot adversarial prompt framework where adapting input sequences with limited data makes significant adversarial robustness improvement. Specifically, we achieve this by providing adversarially correlated text supervision that is end-to-end learned from adversarial examples. We also propose a novel training objective that enhances the consistency of multi-modal features while encourages differentiated uni-modal features between natural and adversarial examples. The proposed framework gives access to learn adversarial text supervision, which provides superior cross-modal adversarial alignment and matches state-of-the-art zero-shot adversarial robustness with only 1% training data. Code is available at: https://github.com/lionel-w2/FAP.

comment: NeurIPS 2024

♻ ☆ Reducing Variance in Meta-Learning via Laplace Approximation for Regression Tasks

Alfredo Reichlin, Gustaf Tegnér, Miguel Vasco, Hang Yin, Mårten Björkman, Danica Kragic

Given a finite set of sample points, meta-learning algorithms aim to learn an optimal adaptation strategy for new, unseen tasks. Often, this data can be ambiguous as it might belong to different tasks concurrently. This is particularly the case in meta-regression tasks. In such cases, the estimated adaptation strategy is subject to high variance due to the limited amount of support data for each task, which often leads to sub-optimal generalization performance. In this work, we address the problem of variance reduction in gradient-based meta-learning and formalize the class of problems prone to this, a condition we refer to as \emph{task overlap}. Specifically, we propose a novel approach that reduces the variance of the gradient estimate by weighing each support point individually by the variance of its posterior over the parameters. To estimate the posterior, we utilize the Laplace approximation, which allows us to express the variance in terms of the curvature of the loss landscape of our meta-learner. Experimental results demonstrate the effectiveness of the proposed method and highlight the importance of variance reduction in meta-learning.

♻ ☆ Conquering the Communication Constraints to Enable Large Pre-Trained Models in Federated Learning

Guangyu Sun, Umar Khalid, Matias Mendieta, Taojiannan Yang, Pu Wang, Minwoo Lee, Chen Chen

Federated learning (FL) has emerged as a promising paradigm for enabling the collaborative training of models without centralized access to the raw data on local devices. In the typical FL paradigm (e.g., FedAvg), model weights are sent to and from the server each round to participating clients. Recently, the use of small pre-trained models has been shown effective in federated learning optimization and improving convergence. However, recent state-of-the-art pre-trained models are getting more capable but also have more parameters. In conventional FL, sharing the enormous model weights can quickly put a massive communication burden on the system, especially if more capable models are employed. Can we find a solution to enable those strong and readily-available pre-trained models in FL to achieve excellent performance while simultaneously reducing the communication burden? To this end, we investigate the use of parameter-efficient fine-tuning in federated learning and thus introduce a new framework: FedPEFT. Specifically, we systemically evaluate the performance of FedPEFT across a variety of client stability, data distribution, and differential privacy settings. By only locally tuning and globally sharing a small portion of the model weights, significant reductions in the total communication overhead can be achieved while maintaining competitive or even better performance in a wide range of federated learning scenarios, providing insight into a new paradigm for practical and effective federated systems.

♻ ☆ Generative AI Models for Different Steps in Architectural Design: A Literature Review

Chengyuan Li, Tianyu Zhang, Xusheng Du, Ye Zhang, Haoran Xie

Recent advances in generative artificial intelligence (AI) technologies have been significantly driven by models such as generative adversarial networks (GANs), variational autoencoders (VAEs), and denoising diffusion probabilistic models (DDPMs). Although architects recognize the potential of generative AI in design, personal barriers often restrict their access to the latest technological developments, thereby causing the application of generative AI in architectural design to lag behind. Therefore, it is essential to comprehend the principles and advancements of generative AI models and analyze their relevance in architecture applications. This paper first provides an overview of generative AI technologies, with a focus on probabilistic diffusion models (DDPMs), 3D generative models, and foundation models, highlighting their recent developments and main application scenarios. Then, the paper explains how the abovementioned models could be utilized in architecture. We subdivide the architectural design process into six steps and review related research projects in each step from 2020 to the present. Lastly, this paper discusses potential future directions for applying generative AI in the architectural design steps. This research can help architects quickly understand the development and latest progress of generative AI and contribute to the further development of intelligent architecture.

comment: 34 pages, 14 figures, accepted by Frontiers of Architectural Research

♻ ☆ Simplifying Deep Temporal Difference Learning

Matteo Gallici, Mattie Fellows, Benjamin Ellis, Bartomeu Pou, Ivan Masmitja, Jakob Nicolaus Foerster, Mario Martin

Q-learning played a foundational role in the field reinforcement learning (RL). However, TD algorithms with off-policy data, such as Q-learning, or nonlinear function approximation like deep neural networks require several additional tricks to stabilise training, primarily a replay buffer and target networks. Unfortunately, the delayed updating of frozen network parameters in the target network harms the sample efficiency and, similarly, the replay buffer introduces memory and implementation overheads. In this paper, we investigate whether it is possible to accelerate and simplify TD training while maintaining its stability. Our key theoretical result demonstrates for the first time that regularisation techniques such as LayerNorm can yield provably convergent TD algorithms without the need for a target network, even with off-policy data. Empirically, we find that online, parallelised sampling enabled by vectorised environments stabilises training without the need of a replay buffer. Motivated by these findings, we propose PQN, our simplified deep online Q-Learning algorithm. Surprisingly, this simple algorithm is competitive with more complex methods like: Rainbow in Atari, R2D2 in Hanabi, QMix in Smax, PPO-RNN in Craftax, and can be up to 50x faster than traditional DQN without sacrificing sample efficiency. In an era where PPO has become the go-to RL algorithm, PQN reestablishes Q-learning as a viable alternative.

♻ ☆ Stable generative modeling using Schrödinger bridges

Georg A. Gottwald, Fengyi Li, Youssef Marzouk, Sebastian Reich

We consider the problem of sampling from an unknown distribution for which only a sufficiently large number of training samples are available. Such settings have recently drawn considerable interest in the context of generative modelling and Bayesian inference. In this paper, we propose a generative model combining Schr\"odinger bridges and Langevin dynamics. Schr\"odinger bridges over an appropriate reversible reference process are used to approximate the conditional transition probability from the available training samples, which is then implemented in a discrete-time reversible Langevin sampler to generate new samples. By setting the kernel bandwidth in the reference process to match the time step size used in the unadjusted Langevin algorithm, our method effectively circumvents any stability issues typically associated with the time-stepping of stiff stochastic differential equations. Moreover, we introduce a novel split-step scheme, ensuring that the generated samples remain within the convex hull of the training samples. Our framework can be naturally extended to generate conditional samples and to Bayesian inference problems. We demonstrate the performance of our proposed scheme through experiments on synthetic datasets with increasing dimensions and on a stochastic subgrid-scale parametrization conditional sampling problem as well as generating sample trajectories of a dynamical system using conditional sampling.

♻ ☆ Causality-Aware Spatiotemporal Graph Neural Networks for Spatiotemporal Time Series Imputation CIKM'2024

Baoyu Jing, Dawei Zhou, Kan Ren, Carl Yang

Spatiotemporal time series are usually collected via monitoring sensors placed at different locations, which usually contain missing values due to various failures, such as mechanical damages and Internet outages. Imputing the missing values is crucial for analyzing time series. When recovering a specific data point, most existing methods consider all the information relevant to that point regardless of the cause-and-effect relationship. During data collection, it is inevitable that some unknown confounders are included, e.g., background noise in time series and non-causal shortcut edges in the constructed sensor network. These confounders could open backdoor paths and establish non-causal correlations between the input and output. Over-exploiting these non-causal correlations could cause overfitting. In this paper, we first revisit spatiotemporal time series imputation from a causal perspective and show how to block the confounders via the frontdoor adjustment. Based on the results of frontdoor adjustment, we introduce a novel Causality-Aware Spatiotemporal Graph Neural Network (Casper), which contains a novel Prompt Based Decoder (PBD) and a Spatiotemporal Causal Attention (SCA). PBD could reduce the impact of confounders and SCA could discover the sparse causal relationships among embeddings. Theoretical analysis reveals that SCA discovers causal relationships based on the values of gradients. We evaluate Casper on three real-world datasets, and the experimental results show that Casper could outperform the baselines and could effectively discover causal relationships.

comment: Accepted by CIKM'2024. Fixed typos

♻ ☆ MoC-System: Efficient Fault Tolerance for Sparse Mixture-of-Experts Model Training

Weilin Cai, Le Qin, Jiayi Huang

As large language models continue to scale up, distributed training systems have expanded beyond 10k nodes, intensifying the importance of fault tolerance. Checkpoint has emerged as the predominant fault tolerance strategy, with extensive studies dedicated to optimizing its efficiency. However, the advent of the sparse Mixture-of-Experts (MoE) model presents new challenges due to the substantial increase in model size, despite comparable computational demands to dense models. In this work, we propose the Mixture-of-Checkpoint System (MoC-System) to orchestrate the vast array of checkpoint shards produced in distributed training systems. MoC-System features a novel Partial Experts Checkpointing (PEC) mechanism, an algorithm-system co-design that strategically saves a selected subset of experts, effectively reducing the MoE checkpoint size to levels comparable with dense models. Incorporating hybrid parallel strategies, MoC-System involves fully sharded checkpointing strategies to evenly distribute the workload across distributed ranks. Furthermore, MoC-System introduces a two-level checkpointing management method that asynchronously handles in-memory snapshots and persistence processes. We build MoC-System upon the Megatron-DeepSpeed framework, achieving up to a 98.9% reduction in overhead for each checkpointing process compared to the original method, during MoE model training with ZeRO-2 data parallelism and expert parallelism. Additionally, extensive empirical analyses substantiate that our methods enhance efficiency while maintaining comparable model accuracy, even achieving an average accuracy increase of 1.08% on downstream tasks.

♻ ☆ Automated Contrastive Learning Strategy Search for Time Series CIKM'2024

Baoyu Jing, Yansen Wang, Guoxin Sui, Jing Hong, Jingrui He, Yuqing Yang, Dongsheng Li, Kan Ren

In recent years, Contrastive Learning (CL) has become a predominant representation learning paradigm for time series. Most existing methods manually build specific CL Strategies (CLS) by human heuristics for certain datasets and tasks. However, manually developing CLS usually requires excessive prior knowledge about the data, and massive experiments to determine the detailed CL configurations. In this paper, we present an Automated Machine Learning (AutoML) practice at Microsoft, which automatically learns CLS for time series datasets and tasks, namely Automated Contrastive Learning (AutoCL). We first construct a principled search space of size over $3\times10^{12}$, covering data augmentation, embedding transformation, contrastive pair construction, and contrastive losses. Further, we introduce an efficient reinforcement learning algorithm, which optimizes CLS from the performance on the validation tasks, to obtain effective CLS within the space. Experimental results on various real-world datasets demonstrate that AutoCL could automatically find the suitable CLS for the given dataset and task. From the candidate CLS found by AutoCL on several public datasets/tasks, we compose a transferable Generally Good Strategy (GGS), which has a strong performance for other datasets. We also provide empirical analysis as a guide for the future design of CLS.

comment: Accepted by CIKM'2024. Fixed typos

♻ ☆ Bayesian Analysis of Combinatorial Gaussian Process Bandits

Jack Sandberg, Niklas Åkerblom, Morteza Haghir Chehreghani

We consider the combinatorial volatile Gaussian process (GP) semi-bandit problem. Each round, an agent is provided a set of available base arms and must select a subset of them to maximize the long-term cumulative reward. We study the Bayesian setting and provide novel Bayesian cumulative regret bounds for three GP-based algorithms: GP-UCB, GP-BayesUCB and GP-TS. Our bounds extend previous results for GP-UCB and GP-TS to the infinite, volatile and combinatorial setting, and to the best of our knowledge, we provide the first regret bound for GP-BayesUCB. Volatile arms encompass other widely considered bandit problems such as contextual bandits. Furthermore, we employ our framework to address the challenging real-world problem of online energy-efficient navigation, where we demonstrate its effectiveness compared to the alternatives.

comment: 32 pages, 10 figures

♻ ☆ On the explainability of quantum neural networks based on variational quantum circuits

Ammar Daskin

Ridge functions are used to describe and study the lower bound of the approximation done by the neural networks which can be written as a linear combination of activation functions. If the activation functions are also ridge functions, these networks are called explainable neural networks. In this brief paper, we first show that quantum neural networks which are based on variational quantum circuits can be written as a linear combination of ridge functions by following matrix notations. Consequently, we show that the interpretability and explainability of such quantum neural networks can be directly considered and studied as an approximation with the linear combination of ridge functions.

comment: a brief paper,a few missing references have been added

♻ ☆ Conformal Prediction for Causal Effects of Continuous Treatments

Maresa Schröder, Dennis Frauen, Jonas Schweisthal, Konstantin Heß, Valentyn Melnychuk, Stefan Feuerriegel

Uncertainty quantification of causal effects is crucial for safety-critical applications such as personalized medicine. A powerful approach for this is conformal prediction, which has several practical benefits due to model-agnostic finite-sample guarantees. Yet, existing methods for conformal prediction of causal effects are limited to binary/discrete treatments and make highly restrictive assumptions such as known propensity scores. In this work, we provide a novel conformal prediction method for potential outcomes of continuous treatments. We account for the additional uncertainty introduced through propensity estimation so that our conformal prediction intervals are valid even if the propensity score is unknown. Our contributions are three-fold: (1) We derive finite-sample prediction intervals for potential outcomes of continuous treatments. (2) We provide an algorithm for calculating the derived intervals. (3) We demonstrate the effectiveness of the conformal prediction intervals in experiments on synthetic and real-world datasets. To the best of our knowledge, we are the first to propose conformal prediction for continuous treatments when the propensity score is unknown and must be estimated from data.

♻ ☆ Over-the-Air Federated Learning in Cell-Free MIMO with Long-term Power Constraint

Yifan Wang, Cheng Zhang, Yuanndon Zhuang, Mingzeng Dai, Haiming Wang, Yongming Huang

Wireless networks supporting artificial intelligence have gained significant attention, with Over-the-Air Federated Learning emerging as a key application due to its unique transmission and distributed computing characteristics. This paper derives error bounds for Over-the-Air Federated Learning in a Cell-free MIMO system and formulates an optimization problem to minimize optimality gap via joint optimization of power control and beamforming. We introduce the MOP-LOFPC algorithm, which employs Lyapunov optimization to decouple long-term constraints across rounds while requiring only causal channel state information. Experimental results demonstrate that MOP-LOFPC achieves a better and more flexible trade-off between the model's training loss and adherence to long-term power constraints compared to existing baselines.

♻ ☆ Improve Value Estimation of Q Function and Reshape Reward with Monte Carlo Tree Search

Jiamian Li

Reinforcement learning has achieved remarkable success in perfect information games such as Go and Atari, enabling agents to compete at the highest levels against human players. However, research in reinforcement learning for imperfect information games has been relatively limited due to the more complex game structures and randomness. Traditional methods face challenges in training and improving performance in imperfect information games due to issues like inaccurate Q value estimation and reward sparsity. In this paper, we focus on Uno, an imperfect information game, and aim to address these problems by reducing Q value overestimation and reshaping reward function. We propose a novel algorithm that utilizes Monte Carlo Tree Search to average the value estimations in Q function. Even though we choose Double Deep Q Learning as the foundational framework in this paper, our method can be generalized and used in any algorithm which needs Q value estimation, such as the Actor-Critic. Additionally, we employ Monte Carlo Tree Search to reshape the reward structure in the game environment. We compare our algorithm with several traditional methods applied to games such as Double Deep Q Learning, Deep Monte Carlo and Neural Fictitious Self Play, and the experiments demonstrate that our algorithm consistently outperforms these approaches, especially as the number of players in Uno increases, indicating a higher level of difficulty.

♻ ☆ Using Stochastic Gradient Descent to Smooth Nonconvex Functions: Analysis of Implicit Graduated Optimization

Naoki Sato, Hideaki Iiduka

The graduated optimization approach is a heuristic method for finding global optimal solutions for nonconvex functions by using a function smoothing operation with stochastic noise. We show that stochastic noise in stochastic gradient descent (SGD) has the effect of smoothing the objective function, the degree of which is determined by the learning rate, batch size, and variance of the stochastic gradient. Using this finding, we propose and analyze a new graduated optimization algorithm that varies the degree of smoothing by varying the learning rate and batch size, and provide experimental results on image classification tasks with ResNets that support our theoretical findings. We further show that there is an interesting correlation between the degree of smoothing by SGD's stochastic noise, the well-studied ``sharpness'' indicator, and the generalization performance of the model.

comment: The latest version was updated in October 2024. Under review

♻ ☆ Do causal predictors generalize better to new domains? NeurIPS'24

Vivian Y. Nastl, Moritz Hardt

We study how well machine learning models trained on causal features generalize across domains. We consider 16 prediction tasks on tabular datasets covering applications in health, employment, education, social benefits, and politics. Each dataset comes with multiple domains, allowing us to test how well a model trained in one domain performs in another. For each prediction task, we select features that have a causal influence on the target of prediction. Our goal is to test the hypothesis that models trained on causal features generalize better across domains. Without exception, we find that predictors using all available features, regardless of causality, have better in-domain and out-of-domain accuracy than predictors using causal features. Moreover, even the absolute drop in accuracy from one domain to the other is no better for causal predictors than for models that use all features. In addition, we show that recent causal machine learning methods for domain generalization do not perform better in our evaluation than standard predictors trained on the set of causal features. Likewise, causal discovery algorithms either fail to run or select causal variables that perform no better than our selection. Extensive robustness checks confirm that our findings are stable under variable misclassification.

comment: 118 pages, 55 figures, accepted at NeurIPS'24

♻ ☆ Explainable Hierarchical Urban Representation Learning for Commuting Flow Prediction

Mingfei Cai, Yanbo Pang, Yoshihide Sekimoto

Commuting flow prediction is an essential task for municipal operations in the real world. Previous studies have revealed that it is feasible to estimate the commuting origin-destination (OD) demand within a city using multiple auxiliary data. However, most existing methods are not suitable to deal with a similar task at a large scale, namely within a prefecture or the whole nation, owing to the increased number of geographical units that need to be maintained. In addition, region representation learning is a universal approach for gaining urban knowledge for diverse metropolitan downstream tasks. Although many researchers have developed comprehensive frameworks to describe urban units from multi-source data, they have not clarified the relationship between the selected geographical elements. Furthermore, metropolitan areas naturally preserve ranked structures, like cities and their inclusive districts, which makes elucidating relations between cross-level urban units necessary. Therefore, we develop a heterogeneous graph-based model to generate meaningful region embeddings at multiple spatial resolutions for predicting different types of inter-level OD flows. To demonstrate the effectiveness of the proposed method, extensive experiments were conducted using real-world aggregated mobile phone datasets collected from Shizuoka Prefecture, Japan. The results indicate that our proposed model outperforms existing models in terms of a uniform urban structure. We extend the understanding of predicted results using reasonable explanations to enhance the credibility of the model.

♻ ☆ Generative Forests NeurIPS'24

Richard Nock, Mathieu Guillame-Bert

We focus on generative AI for a type of data that still represent one of the most prevalent form of data: tabular data. Our paper introduces two key contributions: a new powerful class of forest-based models fit for such tasks and a simple training algorithm with strong convergence guarantees in a boosting model that parallels that of the original weak / strong supervised learning setting. This algorithm can be implemented by a few tweaks to the most popular induction scheme for decision tree induction (i.e. supervised learning) with two classes. Experiments on the quality of generated data display substantial improvements compared to the state of the art. The losses our algorithm minimize and the structure of our models make them practical for related tasks that require fast estimation of a density given a generative model and an observation (even partially specified): such tasks include missing data imputation and density estimation. Additional experiments on these tasks reveal that our models can be notably good contenders to diverse state of the art methods, relying on models as diverse as (or mixing elements of) trees, neural nets, kernels or graphical models.

comment: NeurIPS'24

♻ ☆ Probabilistic ML Verification via Weighted Model Integration

Paolo Morettin, Andrea Passerini, Roberto Sebastiani

In machine learning (ML) verification, the majority of procedures are non-quantitative and therefore cannot be used for verifying probabilistic models, or be applied in domains where hard guarantees are practically unachievable. The probabilistic formal verification (PFV) of ML models is in its infancy, with the existing approaches limited to specific ML models, properties, or both. This contrasts with standard formal methods techniques, whose successful adoption in real-world scenarios is also due to their support for a wide range of properties and diverse systems. We propose a unifying framework for the PFV of ML systems based on Weighted Model Integration (WMI), a relatively recent formalism for probabilistic inference with algebraic and logical constraints. Crucially, reducing the PFV of ML models to WMI enables the verification of many properties of interest over a wide range of systems, addressing multiple limitations of deterministic verification and ad-hoc algorithms. We substantiate the generality of the approach on prototypical tasks involving the verification of group fairness, monotonicity, robustness to noise, probabilistic local robustness and equivalence among predictors. We characterize the challenges related to the scalability of the approach and, through our WMI-based perspective, we show how successful scaling techniques in the ML verification literature can be generalized beyond their original scope.

♻ ☆ Multi-Excitation Projective Simulation with a Many-Body Physics Inspired Inductive Bias

Philip A. LeMaitre, Marius Krumm, Hans J. Briegel

With the impressive progress of deep learning, applications relying on machine learning are increasingly being integrated into daily life. However, most deep learning models have an opaque, oracle-like nature making it difficult to interpret and understand their decisions. This problem led to the development of the field known as eXplainable Artificial Intelligence (XAI). One method in this field known as Projective Simulation (PS) models a chain-of-thought as a random walk of a particle on a graph with vertices that have concepts attached to them. While this description has various benefits, including the possibility of quantization, it cannot be naturally used to model thoughts that combine several concepts simultaneously. To overcome this limitation, we introduce Multi-Excitation Projective Simulation (mePS), a generalization that considers a chain-of-thought to be a random walk of several particles on a hypergraph. A definition for a dynamic hypergraph is put forward to describe the agent's training history along with applications to AI and hypergraph visualization. An inductive bias inspired by the remarkably successful few-body interaction models used in quantum many-body physics is formalized for our classical mePS framework and employed to tackle the exponential complexity associated with naive implementations of hypergraphs. We prove that our inductive bias reduces the complexity from exponential to polynomial, with the exponent representing the cutoff on how many particles can interact. We numerically apply our method to two toy environments and a more complex scenario modelling the diagnosis of a broken computer. These environments demonstrate the resource savings provided by an appropriate choice of inductive bias, as well as showcasing aspects of interpretability. A quantum model for mePS is also briefly outlined and some future directions for it are discussed.

comment: 26 pages, 8 figures; Code repository at https://github.com/MariusKrumm/ManyBodyMEPS. Reorganized main text for better readability

♻ ☆ Uncertainty Estimation and Quantification for LLMs: A Simple Supervised Approach

Linyu Liu, Yu Pan, Xiaocheng Li, Guanting Chen

In this paper, we study the problem of uncertainty estimation and calibration for LLMs. We begin by formulating the uncertainty estimation problem, a relevant yet underexplored area in existing literature. We then propose a supervised approach that leverages labeled datasets to estimate the uncertainty in LLMs' responses. Based on the formulation, we illustrate the difference between the uncertainty estimation for LLMs and that for standard ML models and explain why the hidden neurons of the LLMs may contain uncertainty information. Our designed approach demonstrates the benefits of utilizing hidden activations to enhance uncertainty estimation across various tasks and shows robust transferability in out-of-distribution settings. We distinguish the uncertainty estimation task from the uncertainty calibration task and show that better uncertainty estimation leads to better calibration performance. Furthermore, our method is easy to implement and adaptable to different levels of model accessibility including black box, grey box, and white box.

comment: 29 pages, 14 figures

♻ ☆ Towards Foundation Model for Chemical Reactor Modeling: Meta-Learning with Physics-Informed Adaptation

Zihao Wang, Zhe Wu

In this work, we present a novel application of foundation models for chemical reactor modeling. Accurate modeling of real-world chemical reactors through first-principles is often challenging, and the process of rebuilding and retraining models for each new chemical process is inefficient. This raises a critical question: can we develop a single, universal neural network (i.e., a foundation model) that can rapidly adapt to any new chemical process in a reactor? To address this, we propose a foundation model for chemical reactor modeling that employs a meta-learning approach, followed by physics-informed fine-tuning on new tasks with only a few data samples. Our model is designed to generalize across three classic reactor types: continuous stirred tank reactors, batch reactors, and plug flow reactors. Compared to conventional methods such as data-driven learning, physics-informed learning, transfer learning, and meta-learning, our approach demonstrates superior performance in few-shot scenarios. Specifically, it shows rapid adaptation to unseen reactions with varying integer orders across different reactor set-ups, requiring minimal data for fine-tuning. Source code is available at https://github.com/killingbear999/chemical-reactor-foundation-model.

♻ ☆ The Art of Imitation: Learning Long-Horizon Manipulation Tasks from Few Demonstrations

Jan Ole von Hartz, Tim Welschehold, Abhinav Valada, Joschka Boedecker

Task Parametrized Gaussian Mixture Models (TP-GMM) are a sample-efficient method for learning object-centric robot manipulation tasks. However, there are several open challenges to applying TP-GMMs in the wild. In this work, we tackle three crucial challenges synergistically. First, end-effector velocities are non-Euclidean and thus hard to model using standard GMMs. We thus propose to factorize the robot's end-effector velocity into its direction and magnitude, and model them using Riemannian GMMs. Second, we leverage the factorized velocities to segment and sequence skills from complex demonstration trajectories. Through the segmentation, we further align skill trajectories and hence leverage time as a powerful inductive bias. Third, we present a method to automatically detect relevant task parameters per skill from visual observations. Our approach enables learning complex manipulation tasks from just five demonstrations while using only RGB-D observations. Extensive experimental evaluations on RLBench demonstrate that our approach achieves state-of-the-art performance with 20-fold improved sample efficiency. Our policies generalize across different environments, object instances, and object positions, while the learned skills are reusable.

♻ ☆ Integral Operator Approaches for Scattered Data Fitting on Spheres

Shao-Bo Lin

This paper focuses on scattered data fitting problems on spheres. We study the approximation performance of a class of weighted spectral filter algorithms, including Tikhonov regularization, Landaweber iteration, spectral cut-off, and iterated Tikhonov, in fitting noisy data with possibly unbounded random noise. For the analysis, we develop an integral operator approach that can be regarded as an extension of the widely used sampling inequality approach and norming set method in the community of scattered data fitting. After providing an equivalence between the operator differences and quadrature rules, we succeed in deriving optimal Sobolev-type error estimates of weighted spectral filter algorithms. Our derived error estimates do not suffer from the saturation phenomenon for Tikhonov regularization in the literature, native-space-barrier for existing error analysis and adapts to different embedding spaces. We also propose a divide-and-conquer scheme to equip weighted spectral filter algorithms to reduce their computational burden and present the optimal approximation error bounds.

♻ ☆ Hadamard Representations: Augmenting Hyperbolic Tangents in RL

Jacob E. Kooi, Mark Hoogendoorn, Vincent François-Lavet

Activation functions are one of the key components of a deep neural network. The most commonly used activation functions can be classed into the category of continuously differentiable (e.g. tanh) and linear-unit functions (e.g. ReLU), both having their own strengths and drawbacks with respect to downstream performance and representation capacity through learning (e.g. measured by the number of dead neurons and the effective rank). In reinforcement learning, the performance of continuously differentiable activations often falls short as compared to linear-unit functions. We provide insights into the vanishing gradients associated with the former, and show that the dying neuron problem is not exclusive to ReLU's. To alleviate vanishing gradients and the resulting dying neuron problem occurring with continuously differentiable activations, we propose a Hadamard representation. Using deep Q-networks and proximal policy optimization in the Atari domain, we show faster learning, a reduction in dead neurons and increased effective rank.

comment: 24 pages, 19 figures, 3 tables

♻ ☆ Binarized Simplicial Convolutional Neural Networks

Yi Yan, Ercan E. Kuruoglu

Graph Neural Networks have a limitation of solely processing features on graph nodes, neglecting data on high-dimensional structures such as edges and triangles. Simplicial Convolutional Neural Networks (SCNN) represent higher-order structures using simplicial complexes to break this limitation albeit still lacking time efficiency. In this paper, we propose a novel neural network architecture on simplicial complexes named Binarized Simplicial Convolutional Neural Networks (Bi-SCNN) based on the combination of simplicial convolution with a binary-sign forward propagation strategy. The usage of the Hodge Laplacian on a binary-sign forward propagation enables Bi-SCNN to efficiently and effectively represent simplicial features that have higher-order structures than traditional graph node representations. Compared to the previous Simplicial Convolutional Neural Networks, the reduced model complexity of Bi-SCNN shortens the execution time without sacrificing the prediction performance and is less prone to the over-smoothing effect. Experimenting with real-world citation and ocean-drifter data confirmed that our proposed Bi-SCNN is efficient and accurate.

♻ ☆ Understanding Gradient Boosting Classifier: Training, Prediction, and the Role of $γ_j$

Hung-Hsuan Chen

The Gradient Boosting Classifier (GBC) is a widely used machine learning algorithm for binary classification, which builds decision trees iteratively to minimize prediction errors. This document explains the GBC's training and prediction processes, focusing on the computation of terminal node values $\gamma_j$, which are crucial to optimizing the logistic loss function. We derive $\gamma_j$ through a Taylor series approximation and provide a step-by-step pseudocode for the algorithm's implementation. The guide explains the theory of GBC and its practical application, demonstrating its effectiveness in binary classification tasks. We provide a step-by-step example in the appendix to help readers understand.

♻ ☆ Statistical Efficiency of Distributional Temporal Difference Learning NeurIPS 2024

Yang Peng, Liangyu Zhang, Zhihua Zhang

Distributional reinforcement learning (DRL) has achieved empirical success in various domains. One core task in the field of DRL is distributional policy evaluation, which involves estimating the return distribution $\eta^\pi$ for a given policy $\pi$. The distributional temporal difference learning has been accordingly proposed, which is an extension of the temporal difference learning (TD) in the classic RL area. In the tabular case, \citet{rowland2018analysis} and \citet{rowland2023analysis} proved the asymptotic convergence of two instances of distributional TD, namely categorical temporal difference learning (CTD) and quantile temporal difference learning (QTD), respectively. In this paper, we go a step further and analyze the finite-sample performance of distributional TD. To facilitate theoretical analysis, we propose non-parametric distributional TD learning (NTD). For a $\gamma$-discounted infinite-horizon tabular Markov decision process, we show that for NTD we need $\tilde{O}\left(\frac{1}{\varepsilon^{2p}(1-\gamma)^{2p+1}}\right)$ iterations to achieve an $\varepsilon$-optimal estimator with high probability, when the estimation error is measured by the $p$-Wasserstein distance. This sample complexity bound is minimax optimal up to logarithmic factors in the case of the $1$-Wasserstein distance. To achieve this, we establish a novel Freedman's inequality in Hilbert spaces, which would be of independent interest. In addition, we revisit CTD, showing that the same non-asymptotic convergence bounds hold for CTD in the case of the $p$-Wasserstein distance for $p\geq 1$.

comment: NeurIPS 2024 (oral)

♻ ☆ On the Design and Performance of Machine Learning Based Error Correcting Decoders

Yuncheng Yuan, Péter Scheepers, Lydia Tasiou, Yunus Can Gültekin, Federico Corradi, Alex Alvarado

This paper analyzes the design and competitiveness of four neural network (NN) architectures recently proposed as decoders for forward error correction (FEC) codes. We first consider the so-called single-label neural network (SLNN) and the multi-label neural network (MLNN) decoders which have been reported to achieve near maximum likelihood (ML) performance. Here, we show analytically that SLNN and MLNN decoders can always achieve ML performance, regardless of the code dimensions -- although at the cost of computational complexity -- and no training is in fact required. We then turn our attention to two transformer-based decoders: the error correction code transformer (ECCT) and the cross-attention message passing transformer (CrossMPT). We compare their performance against traditional decoders, and show that ordered statistics decoding outperforms these transformer-based decoders. The results in this paper cast serious doubts on the application of NN-based FEC decoders in the short and medium block length regime.

comment: 6 pages, 4 figures, submitted for possible presentation in a conference (v2: Pre-FEC BER curves are corrected)

♻ ☆ Adaptive Spatio-temporal Estimation on the Graph Edges via Line Graph Transformation

Yi Yan, Ercan Engin Kuruoglu

Spatio-temporal estimation of signals on graph edges is challenging because most conventional Graph Signal Processing techniques are defined on the graph nodes. Leveraging the Line Graph transform, the Line Graph Least Mean Square (LGLMS) algorithm is proposed to conduct adaptive estimation of time-varying edge signals by projecting the edge signals from edge space to node space. LGLMS is an adaptive algorithm analogous to the classical LMS algorithm but applied to graph edges. Unlike edge-specific methods, LGLMS retains all GSP concepts and techniques originally designed for graph nodes, without the need for redefinition on the edges. Experimenting with transportation graphs and meteorological graphs, with the signal observations having noisy and missing values, we confirmed that LGLMS is suitable for the online prediction of time-varying edge signals.

♻ ☆ Universal approximation results for neural networks with non-polynomial activation function over non-compact domains

Ariel Neufeld, Philipp Schmocker

In this paper, we generalize the universal approximation property of single-hidden-layer feed-forward neural networks beyond the classical formulation over compact domains. More precisely, by assuming that the activation function is non-polynomial, we derive universal approximation results for neural networks within function spaces over non-compact subsets of a Euclidean space, e.g., weighted spaces, $L^p$-spaces, and (weighted) Sobolev spaces over unbounded domains, where the latter includes the approximation of the (weak) derivatives. Furthermore, we provide some dimension-independent rates for approximating a function with sufficiently regular and integrable Fourier transform by neural networks with non-polynomial activation function.

comment: arXiv admin note: text overlap with arXiv:2312.08410

♻ ☆ Generative AI Security: Challenges and Countermeasures

Banghua Zhu, Norman Mu, Jiantao Jiao, David Wagner

Generative AI's expanding footprint across numerous industries has led to both excitement and increased scrutiny. This paper delves into the unique security challenges posed by Generative AI, and outlines potential research directions for managing these risks.

♻ ☆ Empirical investigation of multi-source cross-validation in clinical ECG classification

Tuija Leinonen, David Wong, Antti Vasankari, Ali Wahab, Ramesh Nadarajah, Matti Kaisti, Antti Airola

Traditionally, machine learning-based clinical prediction models have been trained and evaluated on patient data from a single source, such as a hospital. Cross-validation methods can be used to estimate the accuracy of such models on new patients originating from the same source, by repeated random splitting of the data. However, such estimates tend to be highly overoptimistic when compared to accuracy obtained from deploying models to sources not represented in the dataset, such as a new hospital. The increasing availability of multi-source medical datasets provides new opportunities for obtaining more comprehensive and realistic evaluations of expected accuracy through source-level cross-validation designs. In this study, we present a systematic empirical evaluation of standard K-fold cross-validation and leave-source-out cross-validation methods in a multi-source setting. We consider the task of electrocardiogram based cardiovascular disease classification, combining and harmonizing the openly available PhysioNet CinC Challenge 2021 and the Shandong Provincial Hospital datasets for our study. Our results show that K-fold cross-validation, both on single-source and multi-source data, systemically overestimates prediction performance when the end goal is to generalize to new sources. Leave-source-out cross-validation provides more reliable performance estimates, having close to zero bias though larger variability. The evaluation highlights the dangers of obtaining misleading cross-validation results on medical data and demonstrates how these issues can be mitigated when having access to multi-source data.

comment: 19 pages, 5 figures

♻ ☆ P1-KAN an effective Kolmogorov Arnold Network for function approximation

Xavier Warin

A new Kolmogorov-Arnold network (KAN) is proposed to approximate potentially irregular functions in high dimension. We show that it outperforms multilayer perceptrons in terms of accuracy and converges faster. We also compare it with several proposed KAN networks: the original spline-based KAN network appears to be more effective for smooth functions, while the P1-KAN network is more effective for irregular functions.

♻ ☆ Advancing Open-Set Domain Generalization Using Evidential Bi-Level Hardest Domain Scheduler NeurIPS 2024

Kunyu Peng, Di Wen, Kailun Yang, Ao Luo, Yufan Chen, Jia Fu, M. Saquib Sarfraz, Alina Roitberg, Rainer Stiefelhagen

In Open-Set Domain Generalization (OSDG), the model is exposed to both new variations of data appearance (domains) and open-set conditions, where both known and novel categories are present at test time. The challenges of this task arise from the dual need to generalize across diverse domains and accurately quantify category novelty, which is critical for applications in dynamic environments. Recently, meta-learning techniques have demonstrated superior results in OSDG, effectively orchestrating the meta-train and -test tasks by employing varied random categories and predefined domain partition strategies. These approaches prioritize a well-designed training schedule over traditional methods that focus primarily on data augmentation and the enhancement of discriminative feature learning. The prevailing meta-learning models in OSDG typically utilize a predefined sequential domain scheduler to structure data partitions. However, a crucial aspect that remains inadequately explored is the influence brought by strategies of domain schedulers during training. In this paper, we observe that an adaptive domain scheduler benefits more in OSDG compared with prefixed sequential and random domain schedulers. We propose the Evidential Bi-Level Hardest Domain Scheduler (EBiL-HaDS) to achieve an adaptive domain scheduler. This method strategically sequences domains by assessing their reliabilities in utilizing a follower network, trained with confidence scores learned in an evidential manner, regularized by max rebiasing discrepancy, and optimized in a bi-level manner. The results show that our method substantially improves OSDG performance and achieves more discriminative embeddings for both the seen and unseen categories. The source code is publicly available at https://github.com/KPeng9510/EBiL-HaDS.

comment: Accepted to NeurIPS 2024. The source code is publicly available at https://github.com/KPeng9510/EBiL-HaDS

♻ ☆ UCB Exploration for Fixed-Budget Bayesian Best Arm Identification

Rong J. B. Zhu, Yanqi Qiu

We study best-arm identification (BAI) in the fixed-budget setting. Adaptive allocations based on upper confidence bounds (UCBs), such as UCBE, are known to work well in BAI. However, it is well-known that its optimal regret is theoretically dependent on instances, which we show to be an artifact in many fixed-budget BAI problems. In this paper we propose an UCB exploration algorithm that is both theoretically and empirically efficient for the fixed budget BAI problem under a Bayesian setting. The key idea is to learn prior information, which can enhance the performance of UCB-based BAI algorithm as it has done in the cumulative regret minimization problem. We establish bounds on the failure probability and the simple regret for the Bayesian BAI problem, providing upper bounds of order $\tilde{O}(\sqrt{K/n})$, up to logarithmic factors, where $n$ represents the budget and $K$ denotes the number of arms. Furthermore, we demonstrate through empirical results that our approach consistently outperforms state-of-the-art baselines.

♻ ☆ Diffusion Models are Certifiably Robust Classifiers NeurIPS 2024

Huanran Chen, Yinpeng Dong, Shitong Shao, Zhongkai Hao, Xiao Yang, Hang Su, Jun Zhu

Generative learning, recognized for its effective modeling of data distributions, offers inherent advantages in handling out-of-distribution instances, especially for enhancing robustness to adversarial attacks. Among these, diffusion classifiers, utilizing powerful diffusion models, have demonstrated superior empirical robustness. However, a comprehensive theoretical understanding of their robustness is still lacking, raising concerns about their vulnerability to stronger future attacks. In this study, we prove that diffusion classifiers possess $O(1)$ Lipschitzness, and establish their certified robustness, demonstrating their inherent resilience. To achieve non-constant Lipschitzness, thereby obtaining much tighter certified robustness, we generalize diffusion classifiers to classify Gaussian-corrupted data. This involves deriving the evidence lower bounds (ELBOs) for these distributions, approximating the likelihood using the ELBO, and calculating classification probabilities via Bayes' theorem. Experimental results show the superior certified robustness of these Noised Diffusion Classifiers (NDCs). Notably, we achieve over 80% and 70% certified robustness on CIFAR-10 under adversarial perturbations with \(\ell_2\) norms less than 0.25 and 0.5, respectively, using a single off-the-shelf diffusion model without any additional data.

comment: Accepted by NeurIPS 2024

♻ ☆ A Simple Baseline for Predicting Events with Auto-Regressive Tabular Transformers

Alex Stein, Samuel Sharpe, Doron Bergman, Senthil Kumar, C. Bayan Bruss, John Dickerson, Tom Goldstein, Micah Goldblum

Many real-world applications of tabular data involve using historic events to predict properties of new ones, for example whether a credit card transaction is fraudulent or what rating a customer will assign a product on a retail platform. Existing approaches to event prediction include costly, brittle, and application-dependent techniques such as time-aware positional embeddings, learned row and field encodings, and oversampling methods for addressing class imbalance. Moreover, these approaches often assume specific use-cases, for example that we know the labels of all historic events or that we only predict a pre-specified label and not the data's features themselves. In this work, we propose a simple but flexible baseline using standard autoregressive LLM-style transformers with elementary positional embeddings and a causal language modeling objective. Our baseline outperforms existing approaches across popular datasets and can be employed for various use-cases. We demonstrate that the same model can predict labels, impute missing values, or model event sequences.

comment: 10 pages, 6 pages of references+appendix

♻ ☆ Deep Learning for Detecting and Early Predicting Chronic Obstructive Pulmonary Disease from Spirogram Time Series

Shuhao Mei, Xin Li, Yuxi Zhou, Jiahao Xu, Yong Zhang, Yuxuan Wan, Shan Cao, Qinghao Zhao, Shijia Geng, Junqing Xie, Shengyong Chen, Shenda Hong

Chronic Obstructive Pulmonary Disease (COPD) is a chronic lung disease that causes airflow obstruction. Current methods can only detect COPD from prominent features in spirogram (Volume-Flow time series) but cannot predict future COPD risk from subtle data patterns. We propose a deep learning-based method, DeepSpiro, for early prediction of future COPD risk. DeepSpiro consists of four key components: SpiroSmoother for stabilizing the Volume-Flow curve, SpiroEncoder for capturing volume evolution through key patches of varying lengths, SpiroExplainer for integrating heterogeneous data and explaining predictions through volume attention, and SpiroPredictor for predicting the disease risk of undiagnosed high-risk patients based on key patch concavity, with prediction horizons of 1, 2, 3, 4, 5 years, or even longer. Evaluated on the UK Biobank dataset, DeepSpiro achieved an AUC of 0.8328 for COPD detection and demonstrated strong predictive performance for future COPD risk (p-value < 0.001). DeepSpiro effectively predicts the long-term progression of the disease.

♻ ☆ Utility Theory of Synthetic Data Generation

Shirong Xu, Will Wei Sun, Guang Cheng

Synthetic data algorithms are widely employed in industries to generate artificial data for downstream learning tasks. While existing research primarily focuses on empirically evaluating utility of synthetic data, its theoretical understanding is largely lacking. This paper bridges the practice-theory gap by establishing relevant utility theory in a statistical learning framework. It considers two utility metrics: generalization and ranking of models trained on synthetic data. The former is defined as the generalization difference between models trained on synthetic and on real data. By deriving analytical bounds for this utility metric, we demonstrate that the synthetic feature distribution does not need to be similar as that of real data for ensuring comparable generalization of synthetic models, provided proper model specifications in downstream learning tasks. The latter utility metric studies the relative performance of models trained on synthetic data. In particular, we discover that the distribution of synthetic data is not necessarily similar as the real one to ensure consistent model comparison. Interestingly, consistent model comparison is still achievable even when synthetic responses are not well generated, as long as downstream models are separable by a generalization gap. Finally, extensive experiments on non-parametric models and deep neural networks have been conducted to validate these theoretical findings.

♻ ☆ Spectraformer: A Unified Random Feature Framework for Transformer

Duke Nguyen, Aditya Joshi, Flora Salim

Linearization of attention using various kernel approximation and kernel learning techniques has shown promise. Past methods use a subset of combinations of component functions and weight matrices within the random features paradigm. We identify the need for a systematic comparison of different combinations of weight matrices and component functions for attention learning in Transformer. In this work, we introduce Spectraformer, a unified framework for approximating and learning the kernel function in linearized attention of the Transformer. We experiment with broad classes of component functions and weight matrices for three textual tasks in the LRA benchmark. Our empirical findings indicate that different kernels are good at different tasks and that kernel choice is fundamental to performant models. Our code is available at: https://github.com/dukenguyenxyz/spectraformer .

♻ ☆ Set-based Meta-Interpolation for Few-Task Meta-Learning

Seanie Lee, Bruno Andreis, Kenji Kawaguchi, Juho Lee, Sung Ju Hwang

Meta-learning approaches enable machine learning systems to adapt to new tasks given few examples by leveraging knowledge from related tasks. However, a large number of meta-training tasks are still required for generalization to unseen tasks during meta-testing, which introduces a critical bottleneck for real-world problems that come with only few tasks, due to various reasons including the difficulty and cost of constructing tasks. Recently, several task augmentation methods have been proposed to tackle this issue using domain-specific knowledge to design augmentation techniques to densify the meta-training task distribution. However, such reliance on domain-specific knowledge renders these methods inapplicable to other domains. While Manifold Mixup based task augmentation methods are domain-agnostic, we empirically find them ineffective on non-image domains. To tackle these limitations, we propose a novel domain-agnostic task augmentation method, Meta-Interpolation, which utilizes expressive neural set functions to densify the meta-training task distribution using bilevel optimization. We empirically validate the efficacy of Meta-Interpolation on eight datasets spanning across various domains such as image classification, molecule property prediction, text classification and speech recognition. Experimentally, we show that Meta-Interpolation consistently outperforms all the relevant baselines. Theoretically, we prove that task interpolation with the set function regularizes the meta-learner to improve generalization.

comment: First two authors contributed equally. Name order decided by a coin toss

♻ ☆ Quantifying the Gain in Weak-to-Strong Generalization NeurIPS 2024

Moses Charikar, Chirag Pabbaraju, Kirankumar Shiragur

Recent advances in large language models have shown capabilities that are extraordinary and near-superhuman. These models operate with such complexity that reliably evaluating and aligning them proves challenging for humans. This leads to the natural question: can guidance from weak models (like humans) adequately direct the capabilities of strong models? In a recent and somewhat surprising work, Burns et al. (2023) empirically demonstrated that when strong models (like GPT-4) are finetuned using labels generated by weak supervisors (like GPT-2), the strong models outperform their weaker counterparts -- a phenomenon they term weak-to-strong generalization. In this work, we present a theoretical framework for understanding weak-to-strong generalization. Specifically, we show that the improvement in performance achieved by strong models over their weaker counterparts is quantified by the misfit error incurred by the strong model on labels generated by the weaker model. Our theory reveals several curious algorithmic insights. For instance, we can predict the amount by which the strong model will improve over the weak model, and also choose among different weak models to train the strong model, based on its misfit error. We validate our theoretical findings through various empirical assessments.

comment: 19 pages; NeurIPS 2024 camera-ready version with additional experiments, references and discussion

♻ ☆ Real-World Robot Applications of Foundation Models: A Review

Kento Kawaharazuka, Tatsuya Matsushima, Andrew Gambardella, Jiaxian Guo, Chris Paxton, Andy Zeng

Recent developments in foundation models, like Large Language Models (LLMs) and Vision-Language Models (VLMs), trained on extensive data, facilitate flexible application across different tasks and modalities. Their impact spans various fields, including healthcare, education, and robotics. This paper provides an overview of the practical application of foundation models in real-world robotics, with a primary emphasis on the replacement of specific components within existing robot systems. The summary encompasses the perspective of input-output relationships in foundation models, as well as their role in perception, motion planning, and control within the field of robotics. This paper concludes with a discussion of future challenges and implications for practical robot applications.

♻ ☆ Audio-visual cross-modality knowledge transfer for machine learning-based in-situ monitoring in laser additive manufacturing

Jiarui Xie, Mutahar Safdar, Lequn Chen, Seung Ki Moon, Yaoyao Fiona Zhao

Various machine learning (ML)-based in-situ monitoring systems have been developed to detect anomalies and defects in laser additive manufacturing (LAM) processes. While multimodal fusion, which integrates data from visual, audio, and other modalities, can improve monitoring performance, it also increases hardware, computational, and operational costs due to the use of multiple sensor types. This paper introduces a cross-modality knowledge transfer (CMKT) methodology for LAM in-situ monitoring, which transfers knowledge from a source modality to a target modality. CMKT enhances the representativeness of the features extracted from the target modality, allowing the removal of source modality sensors during prediction. This paper proposes three CMKT methods: semantic alignment, fully supervised mapping, and semi-supervised mapping. The semantic alignment method establishes a shared encoded space between modalities to facilitate knowledge transfer. It employs a semantic alignment loss to align the distributions of identical groups (e.g., visual and audio defective groups) and a separation loss to distinguish different groups (e.g., visual defective and audio defect-free groups). The two mapping methods transfer knowledge by deriving features from one modality to another using fully supervised and semi-supervised learning approaches. In a case study for LAM in-situ defect detection, the proposed CMKT methods were compared with multimodal audio-visual fusion. The semantic alignment method achieved an accuracy of 98.7% while removing the audio modality during the prediction phase, which is comparable to the 98.2% accuracy obtained through multimodal fusion. Using explainable artificial intelligence, we discovered that semantic alignment CMKT can extract more representative features while reducing noise by leveraging the inherent correlations between modalities.

comment: 45 pages, 17 figures, 6 tables

♻ ☆ The Dark Side of Rich Rewards: Understanding and Mitigating Noise in VLM Rewards

Sukai Huang, Nir Lipovetzky, Trevor Cohn

While Vision-Language Models (VLMs) are increasingly used to generate reward signals for training embodied agents to follow instructions, our research reveals that agents guided by VLM rewards often underperform compared to those employing only intrinsic (exploration-driven) rewards, contradicting expectations set by recent work. We hypothesize that false positive rewards -- instances where unintended trajectories are incorrectly rewarded -- are more detrimental than false negatives. Our analysis confirms this hypothesis, revealing that the widely used cosine similarity metric is prone to false positive reward estimates. To address this, we introduce BiMI ({Bi}nary {M}utual {I}nformation), a novel reward function designed to mitigate noise. BiMI significantly enhances learning efficiency across diverse and challenging embodied navigation environments. Our findings offer a nuanced understanding of how different types of reward noise impact agent learning and highlight the importance of addressing multimodal reward signal noise when training embodied agents

comment: 10 main body pages, 11 appendix pages

♻ ☆ BrainTransformers: SNN-LLM

Zhengzheng Tang, Eva Zhu

This study introduces BrainTransformers, an innovative Large Language Model (LLM) implemented using Spiking Neural Networks (SNN). Our key contributions include: (1) designing SNN-compatible Transformer components such as SNNMatmul, SNNSoftmax, and SNNSiLU; (2) implementing an SNN approximation of the SiLU activation function; and (3) developing a Synapsis module to simulate synaptic plasticity. Our 3-billion parameter model, BrainTransformers-3B-Chat, demonstrates competitive performance across various benchmarks, including MMLU (63.2), BBH (54.1), ARC-C (54.3), and GSM8K (76.3), while potentially offering improved energy efficiency and biological plausibility. The model employs a three-stage training approach, including SNN-specific neuronal synaptic plasticity training. This research opens new avenues for brain-like AI systems in natural language processing and neuromorphic computing. Future work will focus on hardware optimization, developing specialized SNN fine-tuning tools, and exploring practical applications in energy-efficient computing environments.

♻ ☆ TSDS: Data Selection for Task-Specific Model Finetuning

Zifan Liu, Amin Karbasi, Theodoros Rekatsinas

Finetuning foundation models for specific tasks is an emerging paradigm in modern machine learning. The efficacy of task-specific finetuning largely depends on the selection of appropriate training data. We present TSDS (Task-Specific Data Selection), a framework to select data for task-specific model finetuning, guided by a small but representative set of examples from the target task. To do so, we formulate data selection for task-specific finetuning as an optimization problem with a distribution alignment loss based on optimal transport to capture the discrepancy between the selected data and the target distribution. In addition, we add a regularizer to encourage the diversity of the selected data and incorporate kernel density estimation into the regularizer to reduce the negative effects of near-duplicates among the candidate data. We connect our optimization problem to nearest neighbor search and design efficient algorithms to compute the optimal solution based on approximate nearest neighbor search techniques. We evaluate our method on data selection for both continued pretraining and instruction tuning of language models. We show that instruction tuning using data selected by our method with a 1% selection ratio often outperforms using the full dataset and beats the baseline selection methods by 1.5 points in F1 score on average.

comment: 31 pages, 1 figure

♻ ☆ Comparing Quantum Encoding Techniques

Nidhi Munikote

As quantum computers continue to become more capable, the possibilities of their applications increase. For example, quantum techniques are being integrated with classical neural networks to perform machine learning. In order to be used in this way, or for any other widespread use like quantum chemistry simulations or cryptographic applications, classical data must be converted into quantum states through quantum encoding. There are three fundamental encoding methods: basis, amplitude, and rotation, as well as several proposed combinations. This study explores the encoding methods, specifically in the context of hybrid quantum-classical machine learning. Using the QuClassi quantum neural network architecture to perform binary classification of the `3' and `6' digits from the MNIST datasets, this study obtains several metrics such as accuracy, entropy, loss, and resistance to noise, while considering resource usage and computational complexity to compare the three main encoding methods.

♻ ☆ No more hard prompts: SoftSRV prompting for synthetic data generation

Giulia DeSalvo, Jean-Fracois Kagy, Lazaros Karydas, Afshin Rostamizadeh, Sanjiv Kumar

We present a novel soft prompt based framework, SoftSRV, that leverages a frozen pre-trained large language model (LLM) to generate targeted synthetic text sequences. Given a sample from the target distribution, our proposed framework uses data-driven loss minimization to train a parameterized "contextual" soft prompt. This soft prompt is then used to steer the frozen LLM to generate synthetic sequences that are similar to the target distribution. We argue that SoftSRV provides a practical improvement over common hard-prompting approaches that rely on human-curated prompt-templates, which can be idiosyncratic, labor-intensive to craft, and may need to be specialized per domain. We empirically evaluate SoftSRV and hard-prompting baselines by generating synthetic data to fine-tune a small Gemma model on three different domains (coding, math, reasoning). To stress the generality of SoftSRV, we perform these evaluations without any particular specialization of the framework to each domain. We find that SoftSRV significantly improves upon hard-prompting baselines, generating data with superior fine-tuning performance and that better matches the target distribution according to the MAUVE similarity metric.

♻ ☆ RegExplainer: Generating Explanations for Graph Neural Networks in Regression Task NeurIPS 2024

Jiaxing Zhang, Zhuomin Chen, Hao Mei, Dongsheng Luo, Hua Wei

Graph regression is a fundamental task and has received increasing attention in a wide range of graph learning tasks. However, the inference process is often not interpretable. Most existing explanation techniques are limited to understanding GNN behaviors in classification tasks. In this work, we seek an explanation to interpret the graph regression models (XAIG-R). We show that existing methods overlook the distribution shifting and continuously ordered decision boundary, which hinders them away from being applied in the regression tasks. To address these challenges, we propose a novel objective based on the information bottleneck theory and introduce a new mix-up framework, which could support various GNNs in a model-agnostic manner. We further present a contrastive learning strategy to tackle the continuously ordered labels in regression task. To empirically verify the effectiveness of the proposed method, we introduce three benchmark datasets and a real-life dataset for evaluation. Extensive experiments show the effectiveness of the proposed method in interpreting GNN models in regression tasks.

comment: Accepted by NeurIPS 2024

♻ ☆ RoPINN: Region Optimized Physics-Informed Neural Networks

Haixu Wu, Huakun Luo, Yuezhou Ma, Jianmin Wang, Mingsheng Long

Physics-informed neural networks (PINNs) have been widely applied to solve partial differential equations (PDEs) by enforcing outputs and gradients of deep models to satisfy target equations. Due to the limitation of numerical computation, PINNs are conventionally optimized on finite selected points. However, since PDEs are usually defined on continuous domains, solely optimizing models on scattered points may be insufficient to obtain an accurate solution for the whole domain. To mitigate this inherent deficiency of the default scatter-point optimization, this paper proposes and theoretically studies a new training paradigm as region optimization. Concretely, we propose to extend the optimization process of PINNs from isolated points to their continuous neighborhood regions, which can theoretically decrease the generalization error, especially for hidden high-order constraints of PDEs. A practical training algorithm, Region Optimized PINN (RoPINN), is seamlessly derived from this new paradigm, which is implemented by a straightforward but effective Monte Carlo sampling method. By calibrating the sampling process into trust regions, RoPINN finely balances optimization and generalization error. Experimentally, RoPINN consistently boosts the performance of diverse PINNs on a wide range of PDEs without extra backpropagation or gradient calculation. Code is available at this repository: https://github.com/thuml/RoPINN.

♻ ☆ Do LLMs Have Political Correctness? Analyzing Ethical Biases and Jailbreak Vulnerabilities in AI Systems

Isack Lee, Haebin Seong

Although large language models (LLMs) demonstrate impressive proficiency in various tasks, they present potential safety risks, such as `jailbreaks', where malicious inputs can coerce LLMs into generating harmful content. To address these issues, many LLM developers have implemented various safety measures to align these models. This alignment involves several techniques, including data filtering during pre-training, supervised fine-tuning, reinforcement learning from human feedback, and red-teaming exercises. These methods often introduce deliberate and intentional biases similar to Political Correctness (PC) to ensure the ethical behavior of LLMs. In this paper, we delve into the intentional biases injected into LLMs for safety purposes and examine methods to circumvent these safety alignment techniques. Notably, these intentional biases result in a jailbreaking success rate in GPT-4o models that differs by 20% between non-binary and cisgender keywords and by 16% between white and black keywords, even when the other parts of the prompts are identical. We introduce the concept of PCJailbreak, highlighting the inherent risks posed by these safety-induced biases. Additionally, we propose an efficient defense method PCDefense, which prevents jailbreak attempts by injecting defense prompts prior to generation. PCDefense stands as an appealing alternative to Guard Models, such as Llama-Guard, that require additional inference cost after text generation. Our findings emphasize the urgent need for LLM developers to adopt a more responsible approach when designing and implementing safety measures.

♻ ☆ Online Differentially Private Synthetic Data Generation

Yiyun He, Roman Vershynin, Yizhe Zhu

We present a polynomial-time algorithm for online differentially private synthetic data generation. For a data stream within the hypercube $[0,1]^d$ and an infinite time horizon, we develop an online algorithm that generates a differentially private synthetic dataset at each time $t$. This algorithm achieves a near-optimal accuracy bound of $O(\log(t)t^{-1/d})$ for $d\geq 2$ and $O(\log^{4.5}(t)t^{-1})$ for $d=1$ in the 1-Wasserstein distance. This result extends the previous work on the continual release model for counting queries to Lipschitz queries. Compared to the offline case, where the entire dataset is available at once, our approach requires only an extra polylog factor in the accuracy bound.

comment: 21 pages

♻ ☆ Context-Aware Assistant Selection for Improved Inference Acceleration with Large Language Models EMNLP

Jerry Huang, Prasanna Parthasarathi, Mehdi Rezagholizadeh, Sarath Chandar

Despite their widespread adoption, large language models (LLMs) remain prohibitive to use under resource constraints, with their ever growing sizes only increasing the barrier for use. One noted issue is the high latency associated with auto-regressive generation, rendering large LLMs use dependent on advanced computing infrastructure. Assisted decoding, where a smaller draft model guides a larger target model's generation, has helped alleviate this, but remains dependent on alignment between the two models. Thus if the draft model is insufficiently capable on some domain relative to the target model, performance can degrade. Alternatively, one can leverage multiple draft models to better cover the expertise of the target, but when multiple black-box draft models are available, selecting an assistant without details about its construction can be difficult. To better understand this decision making problem, we observe it as a contextual bandit, where a policy must choose a draft model based on a context. We show that even without prior knowledge of the draft models, creating an offline dataset from only outputs of independent draft/target models and training a policy over the alignment of these outputs can accelerate performance on multiple domains provided the candidates are effective. Further results show this to hold on various settings with multiple assisted decoding candidates, highlighting its flexibility and the advantageous role that such decision making can play.

comment: 2024 Conference on Empirical Methods in Natural Language Processing (EMNLP); 14 pages (9 pages main content + references + appendix)

♻ ☆ Regularized Q-learning NeurIPS2024

Han-Dong Lim, Donghwan Lee

Q-learning is widely used algorithm in reinforcement learning community. Under the lookup table setting, its convergence is well established. However, its behavior is known to be unstable with the linear function approximation case. This paper develops a new Q-learning algorithm that converges when linear function approximation is used. We prove that simply adding an appropriate regularization term ensures convergence of the algorithm. We prove its stability using a recent analysis tool based on switching system models. Moreover, we experimentally show that it converges in environments where Q-learning with linear function approximation has known to diverge. We also provide an error bound on the solution where the algorithm converges.

comment: NeurIPS2024

♻ ☆ Hybrid Spatial Representations for Species Distribution Modeling SDM

Shiran Yuan, Hao Zhao

We address an important problem in ecology called Species Distribution Modeling (SDM), whose goal is to predict whether a species exists at a certain position on Earth. In particular, we tackle a challenging version of this task, where we learn from presence-only data in a community-sourced dataset, model a large number of species simultaneously, and do not use any additional environmental information. Previous work has used neural implicit representations to construct models that achieve promising results. However, implicit representations often generate predictions of limited spatial precision. We attribute this limitation to their inherently global formulation and inability to effectively capture local feature variations. This issue is especially pronounced with presence-only data and a large number of species. To address this, we propose a hybrid embedding scheme that combines both implicit and explicit embeddings. Specifically, the explicit embedding is implemented with a multiresolution hashgrid, enabling our models to better capture local information. Experiments demonstrate that our results exceed other works by a large margin on various standard benchmarks, and that the hybrid representation is better than both purely implicit and explicit ones. Qualitative visualizations and comprehensive ablation studies reveal that our hybrid representation successfully addresses the two main challenges. Our code is open-sourced at https://github.com/Shiran-Yuan/HSR-SDM.

comment: Project codebase https://github.com/Shiran-Yuan/HSR-SDM

♻ ☆ Harnessing Your DRAM and SSD for Sustainable and Accessible LLM Inference with Mixed-Precision and Multi-level Caching

Jie Peng, Zhang Cao, Huaizhi Qu, Zhengyu Zhang, Chang Guo, Yanyong Zhang, Zhichao Cao, Tianlong Chen

Although Large Language Models (LLMs) have demonstrated remarkable capabilities, their massive parameter counts and associated extensive computing make LLMs' deployment the main part of carbon emission from nowadays AI applications. Compared to modern GPUs like H$100$, it would be significantly carbon-sustainable if we could leverage old-fashioned GPUs such as M$40$ (as shown in Figure 1, M$40$ only has one third carbon emission of H$100$'s) for LLM servings. However, the limited High Bandwidth Memory (HBM) available on such GPU often cannot support the loading of LLMs due to the gigantic model size and intermediate activation data, making their serving challenging. For instance, a LLaMA2 model with $70$B parameters typically requires $128$GB for inference, which substantially surpasses $24$GB HBM in a $3090$ GPU and remains infeasible even considering the additional $64$GB DRAM. To address this challenge, this paper proposes a mixed-precision with a model modularization algorithm to enable LLM inference on outdated hardware with resource constraints. (The precision denotes the numerical precision like FP16, INT8, INT4) and multi-level caching (M2Cache).) Specifically, our M2Cache first modulizes neurons in LLM and creates their importance ranking. Then, it adopts a dynamic sparse mixed-precision quantization mechanism in weight space to reduce computational demands and communication overhead at each decoding step. It collectively lowers the operational carbon emissions associated with LLM inference. Moreover, M2Cache introduces a three-level cache management system with HBM, DRAM, and SSDs that complements the dynamic sparse mixed-precision inference. To enhance communication efficiency, M2Cache maintains a neuron-level mixed-precision LRU cache in HBM, a larger layer-aware cache in DRAM, and a full model in SSD.

comment: 24 pages, 13 figures

♻ ☆ ODBAE: a high-performance model identifying complex phenotypes in high-dimensional biological datasets

Yafei Shen, Tao Zhang, Zhiwei Liu, Kalliopi Kostelidou, Ying Xu, Ling Yang

Identifying complex phenotypes from high-dimensional biological data is challenging due to the intricate interdependencies among different physiological indicators. Traditional approaches often focus on detecting outliers in single variables, overlooking the broader network of interactions that contribute to phenotype emergence. Here, we introduce ODBAE (Outlier Detection using Balanced Autoencoders), a machine learning method designed to uncover both subtle and extreme outliers by capturing latent relationships among multiple physiological parameters. ODBAE's revised loss function enhances its ability to detect two key types of outliers: influential points (IP), which disrupt latent correlations between dimensions, and high leverage points (HLP), which deviate from the norm but go undetected by traditional autoencoder-based methods. Using data from the International Mouse Phenotyping Consortium (IMPC), we show that ODBAE can identify knockout mice with complex, multi-indicator phenotypes - normal in individual traits, but abnormal when considered together. In addition, this method reveals novel metabolism-related genes and uncovers coordinated abnormalities across metabolic indicators. Our results highlight the utility of ODBAE in detecting joint abnormalities and advancing our understanding of homeostatic perturbations in biological systems.

♻ ☆ On Catastrophic Inheritance of Large Foundation Models

Hao Chen, Bhiksha Raj, Xing Xie, Jindong Wang

Large foundation models (LFMs) are claiming incredible performances. Yet great concerns have been raised about their mythic and uninterpreted potentials not only in machine learning, but also in various other disciplines. In this position paper, we propose to identify a neglected issue deeply rooted in LFMs: Catastrophic Inheritance, describing the weaknesses and limitations inherited from biased large-scale pre-training data to behaviors of LFMs on the downstream tasks, including samples that are corrupted, long-tailed, noisy, out-of-distributed, to name a few. Such inheritance can potentially cause catastrophes to downstream applications, such as bias, lack of generalization, deteriorated performance, security vulnerability, privacy leakage, and value misalignment. We discuss the challenges behind this issue and propose UIM, a framework to Understand the catastrophic inheritance of LFMs from both pre-training and downstream adaptation, Interpret the implications of catastrophic inheritance on downstream tasks, and how to Mitigate it. UIM aims to unite both the machine learning and social sciences communities for more responsible and promising AI development and deployment.

comment: Accepted by DMLR

☆ ALTA: Compiler-Based Analysis of Transformers

Peter Shaw, James Cohan, Jacob Eisenstein, Kenton Lee, Jonathan Berant, Kristina Toutanova

We propose a new programming language called ALTA and a compiler that can map ALTA programs to Transformer weights. ALTA is inspired by RASP, a language proposed by Weiss et al. (2021), and Tracr (Lindner et al., 2023), a compiler from RASP programs to Transformer weights. ALTA complements and extends this prior work, offering the ability to express loops and to compile programs to Universal Transformers, among other advantages. ALTA allows us to constructively show how Transformers can represent length-invariant algorithms for computing parity and addition, as well as a solution to the SCAN benchmark of compositional generalization tasks, without requiring intermediate scratchpad decoding steps. We also propose tools to analyze cases where the expressibility of an algorithm is established, but end-to-end training on a given training set fails to induce behavior consistent with the desired algorithm. To this end, we explore training from ALTA execution traces as a more fine-grained supervision signal. This enables additional experiments and theoretical analyses relating the learnability of various algorithms to data availability and modeling decisions, such as positional encodings. We make the ALTA framework -- language specification, symbolic interpreter, and weight compiler -- available to the community to enable further applications and insights.

☆ TP-Eval: Tap Multimodal LLMs' Potential in Evaluation by Customizing Prompts

Yuxuan Xie, Tianhua Li, Wenqi Shao, Kaipeng Zhang

Recently, multimodal large language models (MLLMs) have received much attention for their impressive capabilities. The evaluation of MLLMs is becoming critical to analyzing attributes of MLLMs and providing valuable insights. However, current benchmarks overlook the problem of prompt sensitivity - minor prompt variations may lead to significant performance fluctuations. Thus, inappropriate prompts may obscure the models' capabilities, underestimating the models' performance. Moreover, different models have different preferences for different prompts, and thus, using the same prompt for all models will cause evaluation bias. This paper analyzes this deficiency in existing benchmarks and further introduces a new evaluation framework named TP-Eval, which introduces a prompt customization method to reduce evaluation biases and tap models' potential. TP-Eval will rewrite the original prompts to different customized prompts for different models. In particular, we propose some well-designed modules for prompt customization tailored to the scenario of MLLM evaluation. Extensive experiments demonstrate the effectiveness of our approach to uncovering models' capabilities, and TP-Eval should benefit the community in developing more comprehensive and convincing MLLM evaluation benchmarks.

☆ CLEAR: Character Unlearning in Textual and Visual Modalities

Alexey Dontsov, Dmitrii Korzh, Alexey Zhavoronkin, Boris Mikheev, Denis Bobkov, Aibek Alanov, Oleg Y. Rogov, Ivan Oseledets, Elena Tutubalina

Machine Unlearning (MU) is critical for enhancing privacy and security in deep learning models, particularly in large multimodal language models (MLLMs), by removing specific private or hazardous information. While MU has made significant progress in textual and visual modalities, multimodal unlearning (MMU) remains significantly underexplored, partially due to the absence of a suitable open-source benchmark. To address this, we introduce CLEAR, a new benchmark designed to evaluate MMU methods. CLEAR contains 200 fictitious individuals and 3,700 images linked with corresponding question-answer pairs, enabling a thorough evaluation across modalities. We assess 10 MU methods, adapting them for MMU, and highlight new challenges specific to multimodal forgetting. We also demonstrate that simple $\ell_1$ regularization on LoRA weights significantly mitigates catastrophic forgetting, preserving model performance on retained data. The dataset is available at https://huggingface.co/datasets/therem/CLEAR

☆ LongRAG: A Dual-Perspective Retrieval-Augmented Generation Paradigm for Long-Context Question Answering EMNLP 2024

Qingfei Zhao, Ruobing Wang, Yukuo Cen, Daren Zha, Shicheng Tan, Yuxiao Dong, Jie Tang

Long-Context Question Answering (LCQA), a challenging task, aims to reason over long-context documents to yield accurate answers to questions. Existing long-context Large Language Models (LLMs) for LCQA often struggle with the "lost in the middle" issue. Retrieval-Augmented Generation (RAG) mitigates this issue by providing external factual evidence. However, its chunking strategy disrupts the global long-context information, and its low-quality retrieval in long contexts hinders LLMs from identifying effective factual details due to substantial noise. To this end, we propose LongRAG, a general, dual-perspective, and robust LLM-based RAG system paradigm for LCQA to enhance RAG's understanding of complex long-context knowledge (i.e., global information and factual details). We design LongRAG as a plug-and-play paradigm, facilitating adaptation to various domains and LLMs. Extensive experiments on three multi-hop datasets demonstrate that LongRAG significantly outperforms long-context LLMs (up by 6.94%), advanced RAG (up by 6.16%), and Vanilla RAG (up by 17.25%). Furthermore, we conduct quantitative ablation studies and multi-dimensional analyses, highlighting the effectiveness of the system's components and fine-tuning strategies. Data and code are available at https://github.com/QingFei1/LongRAG.

comment: EMNLP 2024 Main

☆ Key Algorithms for Keyphrase Generation: Instruction-Based LLMs for Russian Scientific Keyphrases

Anna Glazkova, Dmitry Morozov, Timur Garipov

Keyphrase selection is a challenging task in natural language processing that has a wide range of applications. Adapting existing supervised and unsupervised solutions for the Russian language faces several limitations due to the rich morphology of Russian and the limited number of training datasets available. Recent studies conducted on English texts show that large language models (LLMs) successfully address the task of generating keyphrases. LLMs allow achieving impressive results without task-specific fine-tuning, using text prompts instead. In this work, we access the performance of prompt-based methods for generating keyphrases for Russian scientific abstracts. First, we compare the performance of zero-shot and few-shot prompt-based methods, fine-tuned models, and unsupervised methods. Then we assess strategies for selecting keyphrase examples in a few-shot setting. We present the outcomes of human evaluation of the generated keyphrases and analyze the strengths and weaknesses of the models through expert assessment. Our results suggest that prompt-based methods can outperform common baselines even using simple text prompts.

comment: The 12th International Conference on Analysis of Images, Social Networks and Texts (AIST'2024)

☆ MiLoRA: Efficient Mixture of Low-Rank Adaptation for Large Language Models Fine-tuning EMNLP 2024

Jingfan Zhang, Yi Zhao, Dan Chen, Xing Tian, Huanran Zheng, Wei Zhu

Low-rank adaptation (LoRA) and its mixture-of-experts (MOE) variants are highly effective parameter-efficient fine-tuning (PEFT) methods. However, they introduce significant latency in multi-tenant settings due to the LoRA modules and MOE routers added to multiple linear modules in the Transformer layer. To address this issue, we propose Mixture of Low-Rank Adaptation (MiLoRA), a novel and efficient LoRA variant. MiLoRA differs from previous MOE-style LoRA methods by considering each LoRA module as an expert and employing a prompt-aware routing mechanism. This mechanism calculates expert routing results once before generating the first new token and reuses these results for subsequent tokens, reducing latency. Extensive experiments and analysis on commonsense reasoning tasks, math reasoning tasks, and widely used LLM evaluation benchmarks demonstrate that MiLoRA consistently outperforms strong PEFT baselines with comparable tunable parameter budgets. Additionally, MiLoRA significantly reduces latency in multi-tenant settings compared to previous LoRA-based methods.

comment: Accepted by EMNLP 2024 Findings. arXiv admin note: substantial text overlap with arXiv:2405.18203

☆ GraphTeam: Facilitating Large Language Model-based Graph Analysis via Multi-Agent Collaboration

Xin Li, Qizhi Chu, Yubin Chen, Yang Liu, Yaoqi Liu, Zekai Yu, Weize Chen, Chen Qian, Chuan Shi, Cheng Yang

Graphs are widely used for modeling relational data in real-world scenarios, such as social networks and urban computing. Existing LLM-based graph analysis approaches either integrate graph neural networks (GNNs) for specific machine learning tasks, limiting their transferability, or rely solely on LLMs' internal reasoning ability, resulting in suboptimal performance. To address these limitations, we take advantage of recent advances in LLM-based agents, which have shown capabilities of utilizing external knowledge or tools for problem solving. By simulating human problem-solving strategies such as analogy and collaboration, we propose a multi-agent system based on LLMs named GraphTeam, for graph analysis. GraphTeam consists of five LLM-based agents from three modules, and the agents with different specialities can collaborate with each other to address complex problems. Specifically, (1) input-output normalization module: the question agent extracts and refines four key arguments from the original question, facilitating the problem understanding, and the answer agent organizes the results to meet the output requirement; (2) external knowledge retrieval module: we first build a knowledge base consisting of relevant documentation and experience information, and then the search agent retrieves the most relevant entries for each question. (3) problem-solving module: given the retrieved information from search agent, the coding agent uses established algorithms via programming to generate solutions, and in case the coding agent does not work, the reasoning agent will directly compute the results without programming. Extensive experiments on six graph analysis benchmarks demonstrate that GraphTeam achieves state-of-the-art performance with an average 25.85% improvement over the best baseline in terms of accuracy. The code and data are available at https://github.com/BUPT-GAMMA/GraphTeam.

☆ Cross-lingual Transfer of Reward Models in Multilingual Alignment

Jiwoo Hong, Noah Lee, Rodrigo Martínez-Castaño, César Rodríguez, James Thorne

Reinforcement learning with human feedback (RLHF) is shown to largely benefit from precise reward models (RMs). However, recent studies in reward modeling schemes are skewed towards English, limiting the applicability of RLHF in multilingual alignments. In this work, we investigate the cross-lingual transfer of RMs trained in diverse languages, primarily from English. Our experimental results demonstrate the strong cross-lingual transfer of English RMs, exceeding target language RMs by 3~4% average increase in Multilingual RewardBench. Furthermore, we analyze the cross-lingual transfer of RMs through the representation shifts. Finally, we perform multilingual alignment to exemplify how cross-lingual transfer in RM propagates to enhanced multilingual instruction-following capability, along with extensive analyses on off-the-shelf RMs. We release the code, model, and data.

☆ Together We Can: Multilingual Automatic Post-Editing for Low-Resource Languages EMNLP 2024

Sourabh Deoghare, Diptesh Kanojia, Pushpak Bhattacharyya

This exploratory study investigates the potential of multilingual Automatic Post-Editing (APE) systems to enhance the quality of machine translations for low-resource Indo-Aryan languages. Focusing on two closely related language pairs, English-Marathi and English-Hindi, we exploit the linguistic similarities to develop a robust multilingual APE model. To facilitate cross-linguistic transfer, we generate synthetic Hindi-Marathi and Marathi-Hindi APE triplets. Additionally, we incorporate a Quality Estimation (QE)-APE multi-task learning framework. While the experimental results underline the complementary nature of APE and QE, we also observe that QE-APE multitask learning facilitates effective domain adaptation. Our experiments demonstrate that the multilingual APE models outperform their corresponding English-Hindi and English-Marathi single-pair models by $2.5$ and $2.39$ TER points, respectively, with further notable improvements over the multilingual APE model observed through multi-task learning ($+1.29$ and $+1.44$ TER points), data augmentation ($+0.53$ and $+0.45$ TER points) and domain adaptation ($+0.35$ and $+0.45$ TER points). We release the synthetic data, code, and models accrued during this study publicly at https://github.com/cfiltnlp/Multilingual-APE.

comment: Accepted at Findings of EMNLP 2024

☆ Dependency Graph Parsing as Sequence Labeling EMNLP-2024

Ana Ezquerro, David Vilares, Carlos Gómez-Rodríguez

Various linearizations have been proposed to cast syntactic dependency parsing as sequence labeling. However, these approaches do not support more complex graph-based representations, such as semantic dependencies or enhanced universal dependencies, as they cannot handle reentrancy or cycles. By extending them, we define a range of unbounded and bounded linearizations that can be used to cast graph parsing as a tagging task, enlarging the toolbox of problems that can be solved under this paradigm. Experimental results on semantic dependency and enhanced UD parsing show that with a good choice of encoding, sequence-labeling dependency graph parsers combine high efficiency with accuracies close to the state of the art, in spite of their simplicity.

comment: Accepted at EMNLP-2024

☆ A Time-Aware Approach to Early Detection of Anorexia: UNSL at eRisk 2024

Horacio Thompson, Marcelo Errecalde

The eRisk laboratory aims to address issues related to early risk detection on the Web. In this year's edition, three tasks were proposed, where Task 2 was about early detection of signs of anorexia. Early risk detection is a problem where precision and speed are two crucial objectives. Our research group solved Task 2 by defining a CPI+DMC approach, addressing both objectives independently, and a time-aware approach, where precision and speed are considered a combined single-objective. We implemented the last approach by explicitly integrating time during the learning process, considering the ERDE{\theta} metric as the training objective. It also allowed us to incorporate temporal metrics to validate and select the optimal models. We achieved outstanding results for the ERDE50 metric and ranking-based metrics, demonstrating consistency in solving ERD problems.

comment: In Conference and Labs of the Evaluation Forum (CLEF 2024), Grenoble, France

☆ Zeitenwenden: Detecting changes in the German political discourse

Kai-Robin Lange, Jonas Rieger, Niklas Benner, Carsten Jentsch

From a monarchy to a democracy, to a dictatorship and back to a democracy -- the German political landscape has been constantly changing ever since the first German national state was formed in 1871. After World War II, the Federal Republic of Germany was formed in 1949. Since then every plenary session of the German Bundestag was logged and even has been digitized over the course of the last few years. We analyze these texts using a time series variant of the topic model LDA to investigate which events had a lasting effect on the political discourse and how the political topics changed over time. This allows us to detect changes in word frequency (and thus key discussion points) in political discourse.

comment: 7 pages, 6 figures

☆ ExpertFlow: Optimized Expert Activation and Token Allocation for Efficient Mixture-of-Experts Inference

Xin He, Shunkang Zhang, Yuxin Wang, Haiyan Yin, Zihao Zeng, Shaohuai Shi, Zhenheng Tang, Xiaowen Chu, Ivor Tsang, Ong Yew Soon

Sparse Mixture of Experts (MoE) models, while outperforming dense Large Language Models (LLMs) in terms of performance, face significant deployment challenges during inference due to their high memory demands. Existing offloading techniques, which involve swapping activated and idle experts between the GPU and CPU, often suffer from rigid expert caching mechanisms. These mechanisms fail to adapt to dynamic routing, leading to inefficient cache utilization, or incur prohibitive costs for prediction training. To tackle these inference-specific challenges, we introduce ExpertFlow, a comprehensive system specifically designed to enhance inference efficiency by accommodating flexible routing and enabling efficient expert scheduling between CPU and GPU. This reduces overhead and boosts system performance. Central to our approach is a predictive routing path-based offloading mechanism that utilizes a lightweight predictor to accurately forecast routing paths before computation begins. This proactive strategy allows for real-time error correction in expert caching, significantly increasing cache hit ratios and reducing the frequency of expert transfers, thereby minimizing I/O overhead. Additionally, we implement a dynamic token scheduling strategy that optimizes MoE inference by rearranging input tokens across different batches. This method not only reduces the number of activated experts per batch but also improves computational efficiency. Our extensive experiments demonstrate that ExpertFlow achieves up to 93.72\% GPU memory savings and enhances inference speed by 2 to 10 times compared to baseline methods, highlighting its effectiveness and utility as a robust solution for resource-constrained inference scenarios.

comment: Mixture-of-Experts, Inference, Offloading

☆ SimRAG: Self-Improving Retrieval-Augmented Generation for Adapting Large Language Models to Specialized Domains

Ran Xu, Hui Liu, Sreyashi Nag, Zhenwei Dai, Yaochen Xie, Xianfeng Tang, Chen Luo, Yang Li, Joyce C. Ho, Carl Yang, Qi He

Retrieval-augmented generation (RAG) enhances the question-answering (QA) abilities of large language models (LLMs) by integrating external knowledge. However, adapting general-purpose RAG systems to specialized fields such as science and medicine poses unique challenges due to distribution shifts and limited access to domain-specific data. To tackle this, we propose SimRAG, a self-training approach that equips the LLM with joint capabilities of question answering and question generation for domain adaptation. Our method first fine-tunes the LLM on instruction-following, question-answering, and search-related data. Then, it prompts the same LLM to generate diverse domain-relevant questions from unlabeled corpora, with an additional filtering strategy to retain high-quality synthetic examples. By leveraging these synthetic examples, the LLM can improve their performance on domain-specific RAG tasks. Experiments on 11 datasets, spanning two backbone sizes and three domains, demonstrate that SimRAG outperforms baselines by 1.2\%--8.6\%.

comment: Work in Progress

☆ ELAICHI: Enhancing Low-resource TTS by Addressing Infrequent and Low-frequency Character Bigrams

Srija Anand, Praveen Srinivasa Varadhan, Mehak Singal, Mitesh M. Khapra

Recent advancements in Text-to-Speech (TTS) technology have led to natural-sounding speech for English, primarily due to the availability of large-scale, high-quality web data. However, many other languages lack access to such resources, relying instead on limited studio-quality data. This scarcity results in synthesized speech that often suffers from intelligibility issues, particularly with low-frequency character bigrams. In this paper, we propose three solutions to address this challenge. First, we leverage high-quality data from linguistically or geographically related languages to improve TTS for the target language. Second, we utilize low-quality Automatic Speech Recognition (ASR) data recorded in non-studio environments, which is refined using denoising and speech enhancement models. Third, we apply knowledge distillation from large-scale models using synthetic data to generate more robust outputs. Our experiments with Hindi demonstrate significant reductions in intelligibility issues, as validated by human evaluators. We propose this methodology as a viable alternative for languages with limited access to high-quality data, enabling them to collectively benefit from shared resources.

comment: 11 pages, 1 figure, 3 tables

☆ Value Residual Learning For Alleviating Attention Concentration In Transformers

Zhanchao Zhou, Tianyi Wu, Zhiyun Jiang, Zhenzhong Lan

Transformers can capture long-range dependencies using self-attention, allowing tokens to attend to all others directly. However, stacking multiple attention layers leads to attention concentration. One natural way to address this issue is to use cross-layer attention, allowing information from earlier layers to be directly accessible to later layers. However, this approach is computationally expensive. To address this problem, we propose Transformer with residual value (ResFormer) which approximates cross-layer attention through adding a residual connection from the values of the the first layer to all subsequent layers. Based on this method, one variant is the Transformer with single layer value (SVFormer), where all layers share the same value embedding from first layer, reducing the KV cache by nearly 50%. Comprehensive empirical evidence demonstrates that ResFormer mitigates attention concentration problem in deeper layers and enhances representation across most layers, outperforming the vanilla Transformer, DenseFormer, and NeuTRENO in training error as well as downstream tasks. SVFormer trains significantly faster than the vanilla Transformer and performs better than other methods like GQA and CLA, with performance influenced by sequence length and cumulative learning rate.

☆ Scaling Diffusion Language Models via Adaptation from Autoregressive Models

Shansan Gong, Shivam Agarwal, Yizhe Zhang, Jiacheng Ye, Lin Zheng, Mukai Li, Chenxin An, Peilin Zhao, Wei Bi, Jiawei Han, Hao Peng, Lingpeng Kong

Diffusion Language Models (DLMs) have emerged as a promising new paradigm for text generative modeling, potentially addressing limitations of autoregressive (AR) models. However, current DLMs have been studied at a smaller scale compared to their AR counterparts and lack fair comparison on language modeling benchmarks. Additionally, training diffusion models from scratch at scale remains challenging. Given the prevalence of open-source AR language models, we propose adapting these models to build text diffusion models. We demonstrate connections between AR and diffusion modeling objectives and introduce a simple continual pre-training approach for training diffusion models. Through systematic evaluation on language modeling, reasoning, and commonsense benchmarks, we show that we can convert AR models ranging from 127M to 7B parameters (GPT2 and LLaMA) into diffusion models DiffuGPT and DiffuLLaMA, using less than 200B tokens for training. Our experimental results reveal that these models outperform earlier DLMs and are competitive with their AR counterparts. We release a suite of DLMs (with 127M, 355M, and 7B parameters) capable of generating fluent text, performing in-context learning, filling in the middle without prompt re-ordering, and following instructions \url{https://github.com/HKUNLP/DiffuLLaMA}.

comment: 25 pages. Code: https://github.com/HKUNLP/DiffuLLaMA

☆ SpeakGer: A meta-data enriched speech corpus of German state and federal parliaments

Kai-Robin Lange, Carsten Jentsch

The application of natural language processing on political texts as well as speeches has become increasingly relevant in political sciences due to the ability to analyze large text corpora which cannot be read by a single person. But such text corpora often lack critical meta information, detailing for instance the party, age or constituency of the speaker, that can be used to provide an analysis tailored to more fine-grained research questions. To enable researchers to answer such questions with quantitative approaches such as natural language processing, we provide the SpeakGer data set, consisting of German parliament debates from all 16 federal states of Germany as well as the German Bundestag from 1947-2023, split into a total of 10,806,105 speeches. This data set includes rich meta data in form of information on both reactions from the audience towards the speech as well as information about the speaker's party, their age, their constituency and their party's political alignment, which enables a deeper analysis. We further provide three exploratory analyses, detailing topic shares of different parties throughout time, a descriptive analysis of the development of the age of an average speaker as well as a sentiment analysis of speeches of different parties with regards to the COVID-19 pandemic.

comment: 10 pages, 3 figures

☆ Understanding Layer Significance in LLM Alignment

Guangyuan Shi, Zexin Lu, Xiaoyu Dong, Wenlong Zhang, Xuanyu Zhang, Yujie Feng, Xiao-Ming Wu

Aligning large language models (LLMs) through fine-tuning is essential for tailoring them to specific applications. Therefore, understanding what LLMs learn during the alignment process is crucial. Recent studies suggest that alignment primarily adjusts a model's presentation style rather than its foundational knowledge, indicating that only certain components of the model are significantly impacted. To delve deeper into LLM alignment, we propose to identify which layers within LLMs are most critical to the alignment process, thereby uncovering how alignment influences model behavior at a granular level. We propose a novel approach to identify the important layers for LLM alignment (ILA). It involves learning a binary mask for each incremental weight matrix in the LoRA algorithm, indicating the significance of each layer. ILA consistently identifies important layers across various alignment datasets, with nearly 90% overlap even with substantial dataset differences, highlighting fundamental patterns in LLM alignment. Experimental results indicate that freezing non-essential layers improves overall model performance, while selectively tuning the most critical layers significantly enhances fine-tuning efficiency with minimal performance loss.

☆ Understanding When Tree of Thoughts Succeeds: Larger Models Excel in Generation, Not Discrimination

Qiqi Chen, Xinpeng Wang, Philipp Mondorf, Michael A. Hedderich, Barbara Plank

Tree of Thoughts (ToT) is a reasoning strategy for Large Language Models (LLMs) that employs a generator to suggest reasoning steps and a discriminator to decide which steps to implement. ToT demonstrates strong performance on reasoning tasks, often surpassing simple methods such as Input-Output (IO) prompting and Chain-of-Thought (CoT) reasoning. However, ToT does not consistently outperform such simpler methods across all models, leaving large knowledge gaps on the conditions under which ToT is most beneficial. In this paper, we analyze the roles of the generator and discriminator separately to better understand the conditions when ToT is beneficial. We find that the generator plays a more critical role than the discriminator in driving the success of ToT. While using even a smaller model as the discriminator, scaling the generator leads to notable improvements in ToT performance, whereas scaling the discriminator with a fixed generator yields only marginal gains. Our results show that models across different scales exhibit comparable discrimination capabilities, yet differ significantly in their generative performance for ToT.

comment: Code: github.com/mainlp/tot-eval

☆ OmniFlatten: An End-to-end GPT Model for Seamless Voice Conversation

Qinglin Zhang, Luyao Cheng, Chong Deng, Qian Chen, Wen Wang, Siqi Zheng, Jiaqing Liu, Hai Yu, Chaohong Tan

Full-duplex spoken dialogue systems significantly advance over traditional turn-based dialogue systems, as they allow simultaneous bidirectional communication, closely mirroring human-human interactions. However, achieving low latency and natural interactions in full-duplex dialogue systems remains a significant challenge, especially considering human conversation dynamics such as interruptions, backchannels, and overlapping speech. In this paper, we introduce a novel End-to-End GPT-based model OmniFlatten for full-duplex conversation, capable of effectively modeling the complex behaviors inherent to natural conversations with low latency. To achieve full-duplex communication capabilities, we propose a multi-stage post-training scheme that progressively adapts a text-based large language model (LLM) backbone into a speech-text dialogue LLM, capable of generating text and speech in real time, without modifying the architecture of the backbone LLM. The training process comprises three stages: modality alignment, half-duplex dialogue learning, and full-duplex dialogue learning. Throughout all training stages, we standardize the data using a flattening operation, which allows us to unify the training methods and the model architecture across different modalities and tasks. Our approach offers a straightforward modeling technique and a promising research direction for developing efficient and natural end-to-end full-duplex spoken dialogue systems. Audio samples of dialogues generated by OmniFlatten can be found at this web site (https://omniflatten.github.io/).

comment: Work in progress

☆ Leveraging the Domain Adaptation of Retrieval Augmented Generation Models for Question Answering and Reducing Hallucination

Salman Rakin, Md. A. R. Shibly, Zahin M. Hossain, Zeeshan Khan, Md. Mostofa Akbar

While ongoing advancements in Large Language Models have demonstrated remarkable success across various NLP tasks, Retrieval Augmented Generation Model stands out to be highly effective on downstream applications like Question Answering. Recently, RAG-end2end model further optimized the architecture and achieved notable performance improvements on domain adaptation. However, the effectiveness of these RAG-based architectures remains relatively unexplored when fine-tuned on specialized domains such as customer service for building a reliable conversational AI system. Furthermore, a critical challenge persists in reducing the occurrence of hallucinations while maintaining high domain-specific accuracy. In this paper, we investigated the performance of diverse RAG and RAG-like architectures through domain adaptation and evaluated their ability to generate accurate and relevant response grounded in the contextual knowledge base. To facilitate the evaluation of the models, we constructed a novel dataset HotelConvQA, sourced from wide range of hotel-related conversations and fine-tuned all the models on our domain specific dataset. We also addressed a critical research gap on determining the impact of domain adaptation on reducing hallucinations across different RAG architectures, an aspect that was not properly measured in prior work. Our evaluation shows positive results in all metrics by employing domain adaptation, demonstrating strong performance on QA tasks and providing insights into their efficacy in reducing hallucinations. Our findings clearly indicate that domain adaptation not only enhances the models' performance on QA tasks but also significantly reduces hallucination across all evaluated RAG architectures.

comment: Initial Version fine-tuned on HotelConvQA

☆ Latent Structures of Intertextuality in French Fiction

Jean Barré

Intertextuality is a key concept in literary theory that challenges traditional notions of text, signification or authorship. It views texts as part of a vast intertextual network that is constantly evolving and being reconfigured. This paper argues that the field of computational literary studies is the ideal place to conduct a study of intertextuality since we have now the ability to systematically compare texts with each others. Specifically, we present a work on a corpus of more than 12.000 French fictions from the 18th, 19th and early 20th century. We focus on evaluating the underlying roles of two literary notions, sub-genres and the literary canon in the framing of textuality. The article attempts to operationalize intertextuality using state-of-the-art contextual language models to encode novels and capture features that go beyond simple lexical or thematic approaches. Previous research (Hughes, 2012) supports the existence of a literary "style of a time", and our findings further reinforce this concept. Our findings also suggest that both subgenres and canonicity play a significant role in shaping textual similarities within French fiction. These discoveries point to the importance of considering genre and canon as dynamic forces that influence the evolution and intertextual connections of literary works within specific historical contexts.

comment: 13 pages, 6 figures. Computational Humanities Research Conference 2024

☆ Local Contrastive Editing of Gender Stereotypes EMNLP 2024

Marlene Lutz, Rochelle Choenni, Markus Strohmaier, Anne Lauscher

Stereotypical bias encoded in language models (LMs) poses a threat to safe language technology, yet our understanding of how bias manifests in the parameters of LMs remains incomplete. We introduce local contrastive editing that enables the localization and editing of a subset of weights in a target model in relation to a reference model. We deploy this approach to identify and modify subsets of weights that are associated with gender stereotypes in LMs. Through a series of experiments, we demonstrate that local contrastive editing can precisely localize and control a small subset (< 0.5%) of weights that encode gender bias. Our work (i) advances our understanding of how stereotypical biases can manifest in the parameter space of LMs and (ii) opens up new avenues for developing parameter-efficient strategies for controlling model properties in a contrastive manner.

comment: Accepted at EMNLP 2024

☆ MojoBench: Language Modeling and Benchmarks for Mojo

Nishat Raihan, Joanna C. S. Santos, Marcos Zampieri

The recently introduced Mojo programming language (PL) by Modular, has received significant attention in the scientific community due to its claimed significant speed boost over Python. Despite advancements in code Large Language Models (LLMs) across various PLs, Mojo remains unexplored in this context. To address this gap, we introduce MojoBench, the first framework for Mojo code generation. MojoBench includes HumanEval-Mojo, a benchmark dataset designed for evaluating code LLMs on Mojo, and Mojo-Coder, the first LLM pretrained and finetuned for Mojo code generation, which supports instructions in 5 natural languages (NLs). Our results show that Mojo-Coder achieves a 30-35% performance improvement over leading models like GPT-4o and Claude-3.5-Sonnet. Furthermore, we provide insights into LLM behavior with underrepresented and unseen PLs, offering potential strategies for enhancing model adaptability. MojoBench contributes to our understanding of LLM capabilities and limitations in emerging programming paradigms fostering more robust code generation systems.

☆ Dialectal and Low Resource Machine Translation for Aromanian COLING 2025

Alexandru-Iulius Jerpelea, Alina-Ştefania Rădoi, Sergiu Nisioi

We present a neural machine translation system that can translate between Romanian, English, and Aromanian (an endangered Eastern Romance language); the first of its kind. BLEU scores range from 17 to 32 depending on the direction and genre of the text. Alongside, we release the biggest known Aromanian-Romanian bilingual corpus, consisting of 79k cleaned sentence pairs. Additional tools such as an agnostic sentence embedder (used for both text mining and automatic evaluation) and a diacritics converter are also presented. We publicly release our findings and models. Finally, we describe the deployment of our quantized model at https://arotranslate.com.

comment: 16 pages, 3 figures, 6 tables, submitted to COLING 2025

☆ CogSteer: Cognition-Inspired Selective Layer Intervention for Efficient Semantic Steering in Large Language Models

Xintong Wang, Jingheng Pan, Longqin Jiang, Liang Ding, Xingshan Li, Chris Biemann

Despite their impressive capabilities, large language models (LLMs) often lack interpretability and can generate toxic content. While using LLMs as foundation models and applying semantic steering methods are widely practiced, we believe that efficient methods should be based on a thorough understanding of LLM behavior. To this end, we propose using eye movement measures to interpret LLM behavior across layers. We find that LLMs exhibit patterns similar to human gaze across layers and different layers function differently. Inspired by these findings, we introduce a heuristic steering layer selection and apply it to layer intervention methods via fine-tuning and inference. Using language toxification and detoxification as test beds, we demonstrate that our proposed CogSteer methods achieve better results in terms of toxicity scores while efficiently saving 97% of the computational resources and 60% of the training time. Our model-agnostic approach can be adopted into various LLMs, contributing to their interpretability and promoting trustworthiness for safe deployment.

☆ Beware of Calibration Data for Pruning Large Language Models

Yixin Ji, Yang Xiang, Juntao Li, Qingrong Xia, Ping Li, Xinyu Duan, Zhefeng Wang, Min Zhang

As large language models (LLMs) are widely applied across various fields, model compression has become increasingly crucial for reducing costs and improving inference efficiency. Post-training pruning is a promising method that does not require resource-intensive iterative training and only needs a small amount of calibration data to assess the importance of parameters. Previous research has primarily focused on designing advanced pruning methods, while different calibration data's impact on pruning performance still lacks systematical exploration. We fill this blank and surprisingly observe that the effects of calibration data even value more than designing advanced pruning strategies, especially for high sparsity. Our preliminary exploration also discloses that using calibration data similar to the training data can yield better performance. As pre-training data is usually inaccessible for advanced LLMs, we further provide a self-generating calibration data synthesis strategy to construct feasible calibration data. We conduct experiments on the recent strong open-source LLMs (e.g., DCLM, and LLaMA-3), and the results show that the proposed method outperforms commonly used calibration data and can effectively enhance strong pruning methods (e.g., Wanda, OWL).

comment: under review

☆ An Adaptive Framework for Generating Systematic Explanatory Answer in Online Q&A Platforms

Ziyang Chen, Xiaobin Wang, Yong Jiang, Jinzhi Liao, Pengjun Xie, Fei Huang, Xiang Zhao

Question Answering (QA) systems face challenges in handling complex questions that require multi-domain knowledge synthesis. The naive RAG models, although effective in information retrieval, struggle with complex questions that require comprehensive and in-depth answers. The pioneering task is defined as explanatory answer generation, which entails handling identified challenges such as the requirement for comprehensive information and logical coherence within the generated context. To address these issues, we refer to systematic thinking theory and propose SynthRAG, an innovative framework designed to enhance QA performance. SynthRAG improves on conventional models by employing adaptive outlines for dynamic content structuring, generating systematic information to ensure detailed coverage, and producing customized answers tailored to specific user inquiries. This structured approach guarantees logical coherence and thorough integration of information, yielding responses that are both insightful and methodically organized. Empirical evaluations underscore SynthRAG's effectiveness, demonstrating its superiority in handling complex questions, overcoming the limitations of naive RAG models, and significantly improving answer quality and depth. Furthermore, an online deployment on the Zhihu platform revealed that SynthRAG's answers achieved notable user engagement, with each response averaging 5.73 upvotes and surpassing the performance of 79.8% of human contributors, highlighting the practical relevance and impact of the proposed framework. Our code is available at https://github.com/czy1999/SynthRAG .

comment: 10 pages, 6 figures

☆ Towards a Similarity-adjusted Surprisal Theory EMNLP 2024

Clara Meister, Mario Giulianelli, Tiago Pimentel

Surprisal theory posits that the cognitive effort required to comprehend a word is determined by its contextual predictability, quantified as surprisal. Traditionally, surprisal theory treats words as distinct entities, overlooking any potential similarity between them. Giulianelli et al. (2023) address this limitation by introducing information value, a measure of predictability designed to account for similarities between communicative units. Our work leverages Ricotta and Szeidl's (2006) diversity index to extend surprisal into a metric that we term similarity-adjusted surprisal, exposing a mathematical relationship between surprisal and information value. Similarity-adjusted surprisal aligns with information value when considering graded similarities and reduces to standard surprisal when words are treated as distinct. Experimental results with reading time data indicate that similarity-adjusted surprisal adds predictive power beyond standard surprisal for certain datasets, suggesting it serves as a complementary measure of comprehension effort.

comment: EMNLP 2024 main conference proceedings

☆ Quantifying the Risks of Tool-assisted Rephrasing to Linguistic Diversity

Mengying Wang, Andreas Spitz

Writing assistants and large language models see widespread use in the creation of text content. While their effectiveness for individual users has been evaluated in the literature, little is known about their proclivity to change language or reduce its richness when adopted by a large user base. In this paper, we take a first step towards quantifying this risk by measuring the semantic and vocabulary change enacted by the use of rephrasing tools on a multi-domain corpus of human-generated text.

☆ ReflecTool: Towards Reflection-Aware Tool-Augmented Clinical Agents

Yusheng Liao, Shuyang Jiang, Yanfeng Wang, Yu Wang

Large Language Models (LLMs) have shown promising potential in the medical domain, assisting with tasks like clinical note generation and patient communication. However, current LLMs are limited to text-based communication, hindering their ability to interact with diverse forms of information in clinical environments. Despite clinical agents succeeding in diverse signal interaction, they are oriented to a single clinical scenario and hence fail for broader applications. To evaluate clinical agents holistically, we propose ClinicalAgent Bench~(CAB), a comprehensive medical agent benchmark consisting of 18 tasks across five key realistic clinical dimensions. Building on this, we introduce ReflecTool, a novel framework that excels at utilizing domain-specific tools within two stages. The first optimization stage progressively enlarges a long-term memory by saving successful solving processes and tool-wise experience of agents in a tiny pre-defined training set. In the following inference stage, ReflecTool can search for supportive successful demonstrations from already built long-term memory to guide the tool selection strategy, and a verifier improves the tool usage according to the tool-wise experience with two verification methods--iterative refinement and candidate selection. Extensive experiments on ClinicalAgent Benchmark demonstrate that ReflecTool surpasses the pure LLMs with more than 10 points and the well-established agent-based methods with 3 points, highlighting its adaptability and effectiveness in solving complex clinical tasks.

comment: 20 pages

☆ Markov Chain of Thought for Efficient Mathematical Reasoning

Wen Yang, Kai Fan, Minpeng Liao

Chain of Thought (CoT) of multi-step benefits from the logical structure of the reasoning steps and task-specific actions, significantly enhancing the mathematical reasoning capabilities of large language models. As the prevalence of long CoT, the number of reasoning steps exceeds manageable token limits and leads to higher computational demands. Inspired by the fundamental logic of human cognition, ``derive, then reduce'', we conceptualize the standard multi-step CoT as a novel Markov Chain of Thought (MCoT). In this study, we consider the mathematical reasoning task, defining each reasoning step as text accompanied by a Python code snippet. To facilitate a longer reasoning path, self-correction is enabled through interactions with the code interpreter. Our MCoT aims to compress previous reasoning steps into a simplified question, enabling efficient next-step inference without relying on a lengthy KV cache. In our experiments, we curate the \texttt{MCoTInstruct} dataset, and the empirical results indicate that MCoT not only significantly enhances efficiency but also maintains comparable accuracy. While much remains to be explored, this work paves the way for exploring the long CoT reasoning abilities of LLMs.

comment: Work in progress

☆ LMLPA: Language Model Linguistic Personality Assessment

Jingyao Zheng, Xian Wang, Simo Hosio, Xiaoxian Xu, Lik-Hang Lee

Large Language Models (LLMs) are increasingly used in everyday life and research. One of the most common use cases is conversational interactions, enabled by the language generation capabilities of LLMs. Just as between two humans, a conversation between an LLM-powered entity and a human depends on the personality of the conversants. However, measuring the personality of a given LLM is currently a challenge. This paper introduces the Language Model Linguistic Personality Assessment (LMLPA), a system designed to evaluate the linguistic personalities of LLMs. Our system helps to understand LLMs' language generation capabilities by quantitatively assessing the distinct personality traits reflected in their linguistic outputs. Unlike traditional human-centric psychometrics, the LMLPA adapts a personality assessment questionnaire, specifically the Big Five Inventory, to align with the operational capabilities of LLMs, and also incorporates the findings from previous language-based personality measurement literature. To mitigate sensitivity to the order of options, our questionnaire is designed to be open-ended, resulting in textual answers. Thus, the AI rater is needed to transform ambiguous personality information from text responses into clear numerical indicators of personality traits. Utilising Principal Component Analysis and reliability validations, our findings demonstrate that LLMs possess distinct personality traits that can be effectively quantified by the LMLPA. This research contributes to Human-Computer Interaction and Human-Centered AI, providing a robust framework for future studies to refine AI personality assessments and expand their applications in multiple areas, including education and manufacturing.

☆ Graphusion: A RAG Framework for Knowledge Graph Construction with a Global Perspective

Rui Yang, Boming Yang, Aosong Feng, Sixun Ouyang, Moritz Blum, Tianwei She, Yuang Jiang, Freddy Lecue, Jinghui Lu, Irene Li

Knowledge Graphs (KGs) are crucial in the field of artificial intelligence and are widely used in downstream tasks, such as question-answering (QA). The construction of KGs typically requires significant effort from domain experts. Large Language Models (LLMs) have recently been used for Knowledge Graph Construction (KGC). However, most existing approaches focus on a local perspective, extracting knowledge triplets from individual sentences or documents, missing a fusion process to combine the knowledge in a global KG. This work introduces Graphusion, a zero-shot KGC framework from free text. It contains three steps: in Step 1, we extract a list of seed entities using topic modeling to guide the final KG includes the most relevant entities; in Step 2, we conduct candidate triplet extraction using LLMs; in Step 3, we design the novel fusion module that provides a global view of the extracted knowledge, incorporating entity merging, conflict resolution, and novel triplet discovery. Results show that Graphusion achieves scores of 2.92 and 2.37 out of 3 for entity extraction and relation recognition, respectively. Moreover, we showcase how Graphusion could be applied to the Natural Language Processing (NLP) domain and validate it in an educational scenario. Specifically, we introduce TutorQA, a new expert-verified benchmark for QA, comprising six tasks and a total of 1,200 QA pairs. Using the Graphusion-constructed KG, we achieve a significant improvement on the benchmark, for example, a 9.2% accuracy improvement on sub-graph completion.

comment: arXiv admin note: substantial text overlap with arXiv:2407.10794

☆ Cross-model Control: Improving Multiple Large Language Models in One-time Training NeurIPS 2024

Jiayi Wu, Hao Sun, Hengyi Cai, Lixin Su, Shuaiqiang Wang, Dawei Yin, Xiang Li, Ming Gao

The number of large language models (LLMs) with varying parameter scales and vocabularies is increasing. While they deliver powerful performance, they also face a set of common optimization needs to meet specific requirements or standards, such as instruction following or avoiding the output of sensitive information from the real world. However, how to reuse the fine-tuning outcomes of one model to other models to reduce training costs remains a challenge. To bridge this gap, we introduce Cross-model Control (CMC), a method that improves multiple LLMs in one-time training with a portable tiny language model. Specifically, we have observed that the logit shift before and after fine-tuning is remarkably similar across different models. Based on this insight, we incorporate a tiny language model with a minimal number of parameters. By training alongside a frozen template LLM, the tiny model gains the capability to alter the logits output by the LLMs. To make this tiny language model applicable to models with different vocabularies, we propose a novel token mapping strategy named PM-MinED. We have conducted extensive experiments on instruction tuning and unlearning tasks, demonstrating the effectiveness of CMC. Our code is available at https://github.com/wujwyi/CMC.

comment: Accepted by NeurIPS 2024

☆ MM-Eval: A Multilingual Meta-Evaluation Benchmark for LLM-as-a-Judge and Reward Models

Guijin Son, Dongkeun Yoon, Juyoung Suk, Javier Aula-Blasco, Mano Aslan, Vu Trong Kim, Shayekh Bin Islam, Jaume Prats-Cristià, Lucía Tormo-Bañuelos, Seungone Kim

Large language models (LLMs) are commonly used as evaluators in tasks (e.g., reward modeling, LLM-as-a-judge), where they act as proxies for human preferences or judgments. This leads to the need for meta-evaluation: evaluating the credibility of LLMs as evaluators. However, existing benchmarks primarily focus on English, offering limited insight into LLMs' effectiveness as evaluators in non-English contexts. To address this, we introduce MM-Eval, a multilingual meta-evaluation benchmark that covers 18 languages across six categories. MM-Eval evaluates various dimensions, including language-specific challenges like linguistics and language hallucinations. Evaluation results show that both proprietary and open-source language models have considerable room for improvement. Further analysis reveals a tendency for these models to assign middle-ground scores to low-resource languages. We publicly release our benchmark and code.

comment: work in progress

☆ Differentially Private Learning Needs Better Model Initialization and Self-Distillation

Ivoline C. Ngong, Joseph P. Near, Niloofar Mireshghallah

Differentially private SGD (DPSGD) enables privacy-preserving training of language models, but often reduces utility, diversity, and linguistic quality. We introduce DPRefine, a three-phase method that initializes a model using data synthesis from a small pre-trained LM with rigorous filtering, applies DP finetuning on private data, and performs self-distillation to refine outputs. This approach significantly outperforms vanilla DPSGD, with AlpacaEval preferring DPRefine's generations in 78.4% of cases across all datasets. Our analysis reveals that DPRefine reduces linguistic errors in generated text by 84.0%, mitigating grammar and spelling errors, commonly associated with DPSGD. It also reduces inconsistencies of non-private models, such as hallucinated details and misattributed quotes. We find that small models like GPT-2 can be effective for initialization and distillation, highlighting their potential in enabling scalable and efficient deployment of privacy-preserving language.

comment: 18 pages

☆ ESpeW: Robust Copyright Protection for LLM-based EaaS via Embedding-Specific Watermark

Zongqi Wang, Baoyuan Wu, Jingyuan Deng, Yujiu Yang

Embeddings as a Service (EaaS) is emerging as a crucial role in AI applications. Unfortunately, EaaS is vulnerable to model extraction attacks, highlighting the urgent need for copyright protection.Although some preliminary works propose applying embedding watermarks to protect EaaS, recent research reveals that these watermarks can be easily removed. Hence, it is crucial to inject robust watermarks resistant to watermark removal attacks.Existing watermarking methods typically inject a target embedding into embeddings through linear interpolation when the text contains triggers. However, this mechanism results in each watermarked embedding having the same component, which makes the watermark easy to identify and eliminate.Motivated by this, in this paper, we propose a novel embedding-specific watermarking (ESpeW) mechanism to offer robust copyright protection for EaaS. Our approach involves injecting unique, yet readily identifiable watermarks into each embedding. Watermarks inserted by ESpeW are designed to maintain a significant distance from one another and to avoid sharing common components, thus making it significantly more challenging to remove the watermarks.Extensive experiments on four popular datasets demonstrate that ESpeW can even watermark successfully against a highly aggressive removal strategy without sacrificing the quality of embeddings.

☆ ProtoLens: Advancing Prototype Learning for Fine-Grained Interpretability in Text Classification

Bowen Wei, Ziwei Zhu

Deep neural networks have achieved remarkable performance in various text-based tasks but often lack interpretability, making them less suitable for applications where transparency is critical. To address this, we propose ProtoLens, a novel prototype-based model that provides fine-grained, sub-sentence level interpretability for text classification. ProtoLens uses a Prototype-aware Span Extraction module to identify relevant text spans associated with learned prototypes and a Prototype Alignment mechanism to ensure prototypes are semantically meaningful throughout training. By aligning the prototype embeddings with human-understandable examples, ProtoLens provides interpretable predictions while maintaining competitive accuracy. Extensive experiments demonstrate that ProtoLens outperforms both prototype-based and non-interpretable baselines on multiple text classification benchmarks. Code and data are available at \url{https://anonymous.4open.science/r/ProtoLens-CE0B/}.

☆ Responsible Multilingual Large Language Models: A Survey of Development, Applications, and Societal Impact

Junhua Liu, Bin Fu

Multilingual Large Language Models (MLLMs) represent a pivotal advancement in democratizing artificial intelligence across linguistic boundaries. While theoretical foundations are well-established, practical implementation guidelines remain scattered. This work bridges this gap by providing a comprehensive end-to-end framework for developing and deploying MLLMs in production environments. We make three distinctive contributions: First, we present an actionable pipeline from data pre-processing through deployment, integrating insights from academic research and industrial applications. Second, using Llama2 as a case study, we provide detailed optimization strategies for enhancing multilingual capabilities, including curriculum learning approaches for balancing high-resource and low-resource languages, tokenization strategies, and effective sampling methods. Third, we offer an interdisciplinary analysis that considers technical, linguistic, and cultural perspectives in MLLM development. Our findings reveal critical challenges in supporting linguistic diversity, with 88.38% of world languages categorized as low-resource, affecting over a billion speakers. We examine practical solutions through real-world applications in customer service, search engines, and machine translation. By synthesizing theoretical frameworks with production-ready implementation strategies, this survey provides essential guidance for practitioners and researchers working to develop more inclusive and effective multilingual AI systems.

☆ Navigate Complex Physical Worlds via Geometrically Constrained LLM

Yongqiang Huang, Wentao Ye, Liyao Li, Junbo Zhao

This study investigates the potential of Large Language Models (LLMs) for reconstructing and constructing the physical world solely based on textual knowledge. It explores the impact of model performance on spatial understanding abilities. To enhance the comprehension of geometric and spatial relationships in the complex physical world, the study introduces a set of geometric conventions and develops a workflow based on multi-layer graphs and multi-agent system frameworks. It examines how LLMs achieve multi-step and multi-objective geometric inference in a spatial environment using multi-layer graphs under unified geometric conventions. Additionally, the study employs a genetic algorithm, inspired by large-scale model knowledge, to solve geometric constraint problems. In summary, this work innovatively explores the feasibility of using text-based LLMs as physical world builders and designs a workflow to enhance their capabilities.

☆ MobileSafetyBench: Evaluating Safety of Autonomous Agents in Mobile Device Control

Juyong Lee, Dongyoon Hahm, June Suk Choi, W. Bradley Knox, Kimin Lee

Autonomous agents powered by large language models (LLMs) show promising potential in assistive tasks across various domains, including mobile device control. As these agents interact directly with personal information and device settings, ensuring their safe and reliable behavior is crucial to prevent undesirable outcomes. However, no benchmark exists for standardized evaluation of the safety of mobile device-control agents. In this work, we introduce MobileSafetyBench, a benchmark designed to evaluate the safety of device-control agents within a realistic mobile environment based on Android emulators. We develop a diverse set of tasks involving interactions with various mobile applications, including messaging and banking applications. To clearly evaluate safety apart from general capabilities, we design separate tasks measuring safety and tasks evaluating helpfulness. The safety tasks challenge agents with managing potential risks prevalent in daily life and include tests to evaluate robustness against indirect prompt injections. Our experiments demonstrate that while baseline agents, based on state-of-the-art LLMs, perform well in executing helpful tasks, they show poor performance in safety tasks. To mitigate these safety concerns, we propose a prompting method that encourages agents to prioritize safety considerations. While this method shows promise in promoting safer behaviors, there is still considerable room for improvement to fully earn user trust. This highlights the urgent need for continued research to develop more robust safety mechanisms in mobile environments. We open-source our benchmark at: https://mobilesafetybench.github.io/.

☆ Large Language Models Still Exhibit Bias in Long Text

Wonje Jeung, Dongjae Jeon, Ashkan Yousefpour, Jonghyun Choi

Existing fairness benchmarks for large language models (LLMs) primarily focus on simple tasks, such as multiple-choice questions, overlooking biases that may arise in more complex scenarios like long-text generation. To address this gap, we introduce the Long Text Fairness Test (LTF-TEST), a framework that evaluates biases in LLMs through essay-style prompts. LTF-TEST covers 14 topics and 10 demographic axes, including gender and race, resulting in 11,948 samples. By assessing both model responses and the reasoning behind them, LTF-TEST uncovers subtle biases that are difficult to detect in simple responses. In our evaluation of five recent LLMs, including GPT-4o and LLaMa3, we identify two key patterns of bias. First, these models frequently favor certain demographic groups in their responses. Second, they show excessive sensitivity toward traditionally disadvantaged groups, often providing overly protective responses while neglecting others. To mitigate these biases, we propose FT-REGARD, a finetuning approach that pairs biased prompts with neutral responses. FT-REGARD reduces gender bias by 34.6% and improves performance by 1.4 percentage points on the BBQ benchmark, offering a promising approach to addressing biases in long-text generation tasks.

comment: 22 page, 38 figures, Neurips (SoLaR Workshop)

☆ Mechanisms of Symbol Processing for In-Context Learning in Transformer Networks

Paul Smolensky, Roland Fernandez, Zhenghao Herbert Zhou, Mattia Opper, Jianfeng Gao

Large Language Models (LLMs) have demonstrated impressive abilities in symbol processing through in-context learning (ICL). This success flies in the face of decades of predictions that artificial neural networks cannot master abstract symbol manipulation. We seek to understand the mechanisms that can enable robust symbol processing in transformer networks, illuminating both the unanticipated success, and the significant limitations, of transformers in symbol processing. Borrowing insights from symbolic AI on the power of Production System architectures, we develop a high-level language, PSL, that allows us to write symbolic programs to do complex, abstract symbol processing, and create compilers that precisely implement PSL programs in transformer networks which are, by construction, 100% mechanistically interpretable. We demonstrate that PSL is Turing Universal, so the work can inform the understanding of transformer ICL in general. The type of transformer architecture that we compile from PSL programs suggests a number of paths for enhancing transformers' capabilities at symbol processing. (Note: The first section of the paper gives an extended synopsis of the entire paper.)

comment: 101 pages (including 30 pages of Appendices), 18 figures

☆ BadFair: Backdoored Fairness Attacks with Group-conditioned Triggers EMNLP 2024

Jiaqi Xue, Qian Lou, Mengxin Zheng

Attacking fairness is crucial because compromised models can introduce biased outcomes, undermining trust and amplifying inequalities in sensitive applications like hiring, healthcare, and law enforcement. This highlights the urgent need to understand how fairness mechanisms can be exploited and to develop defenses that ensure both fairness and robustness. We introduce BadFair, a novel backdoored fairness attack methodology. BadFair stealthily crafts a model that operates with accuracy and fairness under regular conditions but, when activated by certain triggers, discriminates and produces incorrect results for specific groups. This type of attack is particularly stealthy and dangerous, as it circumvents existing fairness detection methods, maintaining an appearance of fairness in normal use. Our findings reveal that BadFair achieves a more than 85% attack success rate in attacks aimed at target groups on average while only incurring a minimal accuracy loss. Moreover, it consistently exhibits a significant discrimination score, distinguishing between pre-defined target and non-target attacked groups across various datasets and models.

comment: Accepted by EMNLP 2024

☆ VoiceTextBlender: Augmenting Large Language Models with Speech Capabilities via Single-Stage Joint Speech-Text Supervised Fine-Tuning

Yifan Peng, Krishna C. Puvvada, Zhehuai Chen, Piotr Zelasko, He Huang, Kunal Dhawan, Ke Hu, Shinji Watanabe, Jagadeesh Balam, Boris Ginsburg

Recent studies have augmented large language models (LLMs) with speech capabilities, leading to the development of speech language models (SpeechLMs). Earlier SpeechLMs focused on single-turn speech-based question answering (QA), where user input comprised a speech context and a text question. More recent studies have extended this to multi-turn conversations, though they often require complex, multi-stage supervised fine-tuning (SFT) with diverse data. Another critical challenge with SpeechLMs is catastrophic forgetting-where models optimized for speech tasks suffer significant degradation in text-only performance. To mitigate these issues, we propose a novel single-stage joint speech-text SFT approach on the low-rank adaptation (LoRA) of the LLM backbone. Our joint SFT combines text-only SFT data with three types of speech-related data: speech recognition and translation, speech-based QA, and mixed-modal SFT. Compared to previous SpeechLMs with 7B or 13B parameters, our 3B model demonstrates superior performance across various speech benchmarks while preserving the original capabilities on text-only tasks. Furthermore, our model shows emergent abilities of effectively handling previously unseen prompts and tasks, including multi-turn, mixed-modal inputs.

☆ Which Client is Reliable?: A Reliable and Personalized Prompt-based Federated Learning for Medical Image Question Answering

He Zhu, Ren Togo, Takahiro Ogawa, Miki Haseyama

Conventional medical artificial intelligence (AI) models face barriers in clinical application and ethical issues owing to their inability to handle the privacy-sensitive characteristics of medical data. We present a novel personalized federated learning (pFL) method for medical visual question answering (VQA) models, addressing privacy reliability challenges in the medical domain. Our method introduces learnable prompts into a Transformer architecture to efficiently train it on diverse medical datasets without massive computational costs. Then we introduce a reliable client VQA model that incorporates Dempster-Shafer evidence theory to quantify uncertainty in predictions, enhancing the model's reliability. Furthermore, we propose a novel inter-client communication mechanism that uses maximum likelihood estimation to balance accuracy and uncertainty, fostering efficient integration of insights across clients.

☆ Is artificial intelligence still intelligence? LLMs generalize to novel adjective-noun pairs, but don't mimic the full human distribution

Hayley Ross, Kathryn Davidson, Najoung Kim

Inferences from adjective-noun combinations like "Is artificial intelligence still intelligence?" provide a good test bed for LLMs' understanding of meaning and compositional generalization capability, since there are many combinations which are novel to both humans and LLMs but nevertheless elicit convergent human judgments. We study a range of LLMs and find that the largest models we tested are able to draw human-like inferences when the inference is determined by context and can generalize to unseen adjective-noun combinations. We also propose three methods to evaluate LLMs on these inferences out of context, where there is a distribution of human-like answers rather than a single correct answer. We find that LLMs show a human-like distribution on at most 75\% of our dataset, which is promising but still leaves room for improvement.

comment: 9 pages (23 pages with appendix). Accepted to GenBench 2024

♻ ☆ Non-myopic Generation of Language Model for Reasoning and Planning

Chang Ma, Haiteng Zhao, Junlei Zhang, Junxian He, Lingpeng Kong

Large Language Models have demonstrated remarkable abilities in reasoning and planning by breaking down complex problems into sequential steps. Despite their success in various domains like mathematical problem-solving and coding, LLMs face challenges in ensuring reliable and optimal planning due to their inherent myopic nature of autoregressive decoding. This paper revisits LLM reasoning from an optimal-control perspective, proposing a novel method, Predictive-Decoding, that leverages Model Predictive Control to enhance planning accuracy. By re-weighting LLM distributions based on foresight trajectories, Predictive-Decoding aims to mitigate early errors and promote non-myopic planning. Our experiments show significant improvements in a wide range of tasks for math, coding, and agents. Furthermore, Predictive-Decoding demonstrates computational efficiency, outperforming search baselines with reduced computational resources. This study provides insights into optimizing LLM planning capabilities.

♻ ☆ LLMScan: Causal Scan for LLM Misbehavior Detection

Mengdi Zhang, Kai Kiat Goh, Peixin Zhang, Jun Sun

Despite the success of Large Language Models (LLMs) across various fields, their potential to generate untruthful, biased and harmful responses poses significant risks, particularly in critical applications. This highlights the urgent need for systematic methods to detect and prevent such misbehavior. While existing approaches target specific issues such as harmful responses, this work introduces LLMScan, an innovative LLM monitoring technique based on causality analysis, offering a comprehensive solution. LLMScan systematically monitors the inner workings of an LLM through the lens of causal inference, operating on the premise that the LLM's `brain' behaves differently when misbehaving. By analyzing the causal contributions of the LLM's input tokens and transformer layers, LLMScan effectively detects misbehavior. Extensive experiments across various tasks and models reveal clear distinctions in the causal distributions between normal behavior and misbehavior, enabling the development of accurate, lightweight detectors for a variety of misbehavior detection tasks.

♻ ☆ Susu Box or Piggy Bank: Assessing Cultural Commonsense Knowledge between Ghana and the U.S EMNLP 2024

Christabel Acquaye, Haozhe An, Rachel Rudinger

Recent work has highlighted the culturally-contingent nature of commonsense knowledge. We introduce AMAMMER${\epsilon}$, a test set of 525 multiple-choice questions designed to evaluate the commonsense knowledge of English LLMs, relative to the cultural contexts of Ghana and the United States. To create AMAMMER${\epsilon}$, we select a set of multiple-choice questions (MCQs) from existing commonsense datasets and rewrite them in a multi-stage process involving surveys of Ghanaian and U.S. participants. In three rounds of surveys, participants from both pools are solicited to (1) write correct and incorrect answer choices, (2) rate individual answer choices on a 5-point Likert scale, and (3) select the best answer choice from the newly-constructed MCQ items, in a final validation step. By engaging participants at multiple stages, our procedure ensures that participant perspectives are incorporated both in the creation and validation of test items, resulting in high levels of agreement within each pool. We evaluate several off-the-shelf English LLMs on AMAMMER${\epsilon}$. Uniformly, models prefer answers choices that align with the preferences of U.S. annotators over Ghanaian annotators. Additionally, when test items specify a cultural context (Ghana or the U.S.), models exhibit some ability to adapt, but performance is consistently better in U.S. contexts than Ghanaian. As large resources are devoted to the advancement of English LLMs, our findings underscore the need for culturally adaptable models and evaluations to meet the needs of diverse English-speaking populations around the world.

comment: Accepted to EMNLP 2024

♻ ☆ 1-bit AI Infra: Part 1.1, Fast and Lossless BitNet b1.58 Inference on CPUs

Jinheng Wang, Hansong Zhou, Ting Song, Shaoguang Mao, Shuming Ma, Hongyu Wang, Yan Xia, Furu Wei

Recent advances in 1-bit Large Language Models (LLMs), such as BitNet and BitNet b1.58, present a promising approach to enhancing the efficiency of LLMs in terms of speed and energy consumption. These developments also enable local LLM deployment across a broad range of devices. In this work, we introduce bitnet.cpp, a tailored software stack designed to unlock the full potential of 1-bit LLMs. Specifically, we develop a set of kernels to support fast and lossless inference of ternary BitNet b1.58 LLMs on CPUs. Extensive experiments demonstrate that bitnet.cpp achieves significant speedups, ranging from 2.37x to 6.17x on x86 CPUs and from 1.37x to 5.07x on ARM CPUs, across various model sizes. The code is available at https://github.com/microsoft/BitNet.

♻ ☆ A Bi-consolidating Model for Joint Relational Triple Extraction

Xiaocheng Luo, Yanping Chen, Ruixue Tang, Caiwei Yang, Ruizhang Huang, Yongbin Qin

Current methods to extract relational triples directly make a prediction based on a possible entity pair in a raw sentence without depending on entity recognition. The task suffers from a serious semantic overlapping problem, in which several relation triples may share one or two entities in a sentence. In this paper, based on a two-dimensional sentence representation, a bi-consolidating model is proposed to address this problem by simultaneously reinforcing the local and global semantic features relevant to a relation triple. This model consists of a local consolidation component and a global consolidation component. The first component uses a pixel difference convolution to enhance semantic information of a possible triple representation from adjacent regions and mitigate noise in neighbouring neighbours. The second component strengthens the triple representation based a channel attention and a spatial attention, which has the advantage to learn remote semantic dependencies in a sentence. They are helpful to improve the performance of both entity identification and relation type classification in relation triple extraction. After evaluated on several publish datasets, the bi-consolidating model achieves competitive performance. Analytical experiments demonstrate the effectiveness of our model for relational triple extraction and give motivation for other natural language processing tasks.

♻ ☆ MADial-Bench: Towards Real-world Evaluation of Memory-Augmented Dialogue Generation NAACL 2025

Junqing He, Liang Zhu, Rui Wang, Xi Wang, Reza Haffari, Jiaxing Zhang

Long-term memory is important for chatbots and dialogue systems (DS) to create consistent and human-like conversations, evidenced by numerous developed memory-augmented DS (MADS). To evaluate the effectiveness of such MADS, existing commonly used evaluation metrics, like retrieval accuracy and perplexity (PPL), mainly focus on query-oriented factualness and language quality assessment. However, these metrics often lack practical value. Moreover, the evaluation dimensions are insufficient for human-like assessment in DS. Regarding memory-recalling paradigms, current evaluation schemes only consider passive memory retrieval while ignoring diverse memory recall with rich triggering factors, e.g., emotions and surroundings, which can be essential in emotional support scenarios. To bridge the gap, we construct a novel Memory-Augmented Dialogue Benchmark (MADail-Bench) covering various memory-recalling paradigms based on cognitive science and psychology theories. The benchmark assesses two tasks separately: memory retrieval and memory recognition with the incorporation of both passive and proactive memory recall data. We introduce new scoring criteria to the evaluation, including memory injection, emotion support (ES) proficiency, and intimacy, to comprehensively assess generated responses. Results from cutting-edge embedding models and large language models on this benchmark indicate the potential for further advancement. Extensive testing further reveals correlations between memory injection, ES proficiency, and intimacy.

comment: Submitted to NAACL 2025

♻ ☆ Does Generative AI speak Nigerian-Pidgin?: Issues about Representativeness and Bias for Multilingualism in LLMs

David Ifeoluwa Adelani, A. Seza Doğruöz, Iyanuoluwa Shode, Anuoluwapo Aremu

Nigeria is a multilingual country with 500+ languages. Naija is a Nigerian-Pidgin spoken by approx. 120M speakers in Nigeria and it is a mixed language (e.g., English, Portuguese, Yoruba, Hausa and Igbo). Although it has mainly been a spoken language until recently, there are now various platforms publishing exclusively in Naija such as Naija Wikipedia. However, it is hard to distinguish by non-native from a larger pidgin languages spoken across West Africa known as West African Pidgin English (WAPE) -- which is more simplied and understandable by wider audience in Ghana, Nigeria, and Cameroon. BBC news platform publishes exclusively in WAPE to cater for several countries in West Africa. In our paper, we show through statistical analyses and Machine Translation experiments that these two creole varieties do not represent each other (i.e., there are linguistic differences in word order and vocabulary) and Generative AI operates only based on WAPE. In other words, Naija is under-represented in Generative AI, and it is hard to teach LLMs with few examples.

comment: under review

♻ ☆ Conditional Language Policy: A General Framework for Steerable Multi-Objective Finetuning EMNLP 2024

Kaiwen Wang, Rahul Kidambi, Ryan Sullivan, Alekh Agarwal, Christoph Dann, Andrea Michi, Marco Gelmi, Yunxuan Li, Raghav Gupta, Avinava Dubey, Alexandre Ramé, Johan Ferret, Geoffrey Cideron, Le Hou, Hongkun Yu, Amr Ahmed, Aranyak Mehta, Léonard Hussenot, Olivier Bachem, Edouard Leurent

Reward-based finetuning is crucial for aligning language policies with intended behaviors (e.g., creativity and safety). A key challenge is to develop steerable language models that trade-off multiple (conflicting) objectives in a flexible and efficient manner. This paper presents Conditional Language Policy (CLP), a general framework for finetuning language models on multiple objectives. Building on techniques from multi-task training and parameter-efficient finetuning, CLP learn steerable models that effectively trade-off conflicting objectives at inference time. Notably, this does not require training or maintaining multiple models to achieve different trade-offs between the objectives. Through extensive experiments and ablations on two summarization datasets, we show that CLP learns steerable language models that outperform and Pareto-dominate the existing approaches for multi-objective finetuning.

comment: 40 pages. Findings of EMNLP 2024

♻ ☆ STAR: SocioTechnical Approach to Red Teaming Language Models

Laura Weidinger, John Mellor, Bernat Guillen Pegueroles, Nahema Marchal, Ravin Kumar, Kristian Lum, Canfer Akbulut, Mark Diaz, Stevie Bergman, Mikel Rodriguez, Verena Rieser, William Isaac

This research introduces STAR, a sociotechnical framework that improves on current best practices for red teaming safety of large language models. STAR makes two key contributions: it enhances steerability by generating parameterised instructions for human red teamers, leading to improved coverage of the risk surface. Parameterised instructions also provide more detailed insights into model failures at no increased cost. Second, STAR improves signal quality by matching demographics to assess harms for specific groups, resulting in more sensitive annotations. STAR further employs a novel step of arbitration to leverage diverse viewpoints and improve label reliability, treating disagreement not as noise but as a valuable contribution to signal quality.

comment: 8 pages, 5 figures, 5 pages appendix. * denotes equal contribution

♻ ☆ Proof of Thought : Neurosymbolic Program Synthesis allows Robust and Interpretable Reasoning NeurIPS 2024

Debargha Ganguly, Srinivasan Iyengar, Vipin Chaudhary, Shivkumar Kalyanaraman

Large Language Models (LLMs) have revolutionized natural language processing, yet they struggle with inconsistent reasoning, particularly in novel domains and complex logical sequences. This research introduces Proof of Thought, a framework that enhances the reliability and transparency of LLM outputs. Our approach bridges LLM-generated ideas with formal logic verification, employing a custom interpreter to convert LLM outputs into First Order Logic constructs for theorem prover scrutiny. Central to our method is an intermediary JSON-based Domain-Specific Language, which by design balances precise logical structures with intuitive human concepts. This hybrid representation enables both rigorous validation and accessible human comprehension of LLM reasoning processes. Key contributions include a robust type system with sort management for enhanced logical integrity, explicit representation of rules for clear distinction between factual and inferential knowledge, and a flexible architecture that allows for easy extension to various domain-specific applications. We demonstrate Proof of Thought's effectiveness through benchmarking on StrategyQA and a novel multimodal reasoning task, showing improved performance in open-ended scenarios. By providing verifiable and interpretable results, our technique addresses critical needs for AI system accountability and sets a foundation for human-in-the-loop oversight in high-stakes domains.

comment: 38th Conference on Neural Information Processing Systems (NeurIPS 2024) System 2 Reasoning At Scale Workshop

♻ ☆ AlleNoise: large-scale text classification benchmark dataset with real-world label noise

Alicja Rączkowska, Aleksandra Osowska-Kurczab, Jacek Szczerbiński, Kalina Jasinska-Kobus, Klaudia Nazarko

Label noise remains a challenge for training robust classification models. Most methods for mitigating label noise have been benchmarked using primarily datasets with synthetic noise. While the need for datasets with realistic noise distribution has partially been addressed by web-scraped benchmarks such as WebVision and Clothing1M, those benchmarks are restricted to the computer vision domain. With the growing importance of Transformer-based models, it is crucial to establish text classification benchmarks for learning with noisy labels. In this paper, we present AlleNoise, a new curated text classification benchmark dataset with real-world instance-dependent label noise, containing over 500,000 examples across approximately 5,600 classes, complemented with a meaningful, hierarchical taxonomy of categories. The noise distribution comes from actual users of a major e-commerce marketplace, so it realistically reflects the semantics of human mistakes. In addition to the noisy labels, we provide human-verified clean labels, which help to get a deeper insight into the noise distribution, unlike web-scraped datasets typically used in the field. We demonstrate that a representative selection of established methods for learning with noisy labels is inadequate to handle such real-world noise. In addition, we show evidence that these algorithms do not alleviate excessive memorization. As such, with AlleNoise, we set the bar high for the development of label noise methods that can handle real-world label noise in text classification tasks. The code and dataset are available for download at https://github.com/allegro/AlleNoise.

♻ ☆ Annotator-Centric Active Learning for Subjective NLP Tasks EMNLP2024

Michiel van der Meer, Neele Falk, Pradeep K. Murukannaiah, Enrico Liscio

Active Learning (AL) addresses the high costs of collecting human annotations by strategically annotating the most informative samples. However, for subjective NLP tasks, incorporating a wide range of perspectives in the annotation process is crucial to capture the variability in human judgments. We introduce Annotator-Centric Active Learning (ACAL), which incorporates an annotator selection strategy following data sampling. Our objective is two-fold: 1) to efficiently approximate the full diversity of human judgments, and 2) to assess model performance using annotator-centric metrics, which value minority and majority perspectives equally. We experiment with multiple annotator selection strategies across seven subjective NLP tasks, employing both traditional and novel, human-centered evaluation metrics. Our findings indicate that ACAL improves data efficiency and excels in annotator-centric performance evaluations. However, its success depends on the availability of a sufficiently large and diverse pool of annotators to sample from.

comment: Accepted at EMNLP2024

♻ ☆ Do Large Language Models Truly Grasp Mathematics? An Empirical Exploration

Wei Xie, Shuoyoucheng Ma, Zhenhua Wang, Enze Wang, Baosheng Wang, Jinshu Su

Despite their proficiency in math tasks, the mechanisms underlying LLMs' mathematical reasoning abilities remain a subject of debate. Recent studies suggest that chain-of-thought (CoT) prompts can bolster mathematical reasoning by encouraging LLMs to employ human-like logical reasoning (System 2), enabling them to excel on the Cognitive Reflection Test (CRT). To assess whether LLMs genuinely possess System 2-like logical reasoning, we introduced targeted modifications to CRT problems. Our findings reveal that, despite the use of CoT prompts, mainstream LLMs, including the latest o1-preview model, continue to exhibit a significant error rate. Further analysis indicates that they predominantly rely on System 1-like intuitive reasoning and pattern matching derived from training data, rather than demonstrating mastery of mathematical thinking. This discovery challenges the prevailing notion that LLMs possess genuine logical reasoning abilities and that CoT can enhance them. Consequently, this work may temper overly optimistic projections regarding LLMs' advancement toward artificial general intelligence.

♻ ☆ Linear Adversarial Concept Erasure ICML 2022

Shauli Ravfogel, Michael Twiton, Yoav Goldberg, Ryan Cotterell

Modern neural models trained on textual data rely on pre-trained representations that emerge without direct supervision. As these representations are increasingly being used in real-world applications, the inability to \emph{control} their content becomes an increasingly important problem. We formulate the problem of identifying and erasing a linear subspace that corresponds to a given concept, in order to prevent linear predictors from recovering the concept. We model this problem as a constrained, linear maximin game, and show that existing solutions are generally not optimal for this task. We derive a closed-form solution for certain objectives, and propose a convex relaxation, \method, that works well for others. When evaluated in the context of binary gender removal, the method recovers a low-dimensional subspace whose removal mitigates bias by intrinsic and extrinsic evaluation. We show that the method is highly expressive, effectively mitigating bias in deep nonlinear classifiers while maintaining tractability and interpretability.

comment: Accepted in ICML 2022; a revised version

♻ ☆ Fast and Slow Generating: An Empirical Study on Large and Small Language Models Collaborative Decoding

Kaiyan Zhang, Jianyu Wang, Ning Ding, Biqing Qi, Ermo Hua, Xingtai Lv, Bowen Zhou

Large Language Models (LLMs) exhibit impressive capabilities across various applications but encounter substantial challenges such as high inference latency, considerable training costs, and the generation of hallucinations. Collaborative decoding between large and small language models (SLMs) presents a promising strategy to mitigate these issues through methods including speculative decoding, contrastive decoding, and emulator or proxy fine-tuning. However, the specifics of such collaborations, particularly from a unified perspective, remain largely unexplored. Inspired by dual-process cognitive theory, we propose a unified framework in this paper, termed Fast and Slow Generating (FS-GEN). Within this framework, LLMs (sometimes along with SLMs) are categorized as System 2 (slow and deliberate), while independent SLMs are designated as System 1 (fast and intuitive). We provide a comprehensive analysis of these collaborative methodologies, elucidating their common properties and shedding light on the differential knowledge capabilities of System 2 versus System 1 through the FS-GEN framework. Our findings indicate that only a small proportion of collaborative interactions (approximately less than 20\% in most instances) are necessary across various methods. These interactions between System 1 and System 2 conform to a scaling law related to the parameter ratios, enabling predictable collaboration. Furthermore, we explore the specific conditions under which collaboration proves most effective, particularly from an uncertainty perspective, offering novel insights that may guide future optimization efforts. Our research underscores that the fundamental distinction between System 1 and System 2 lies in the uncertainty of next token predictions, where interventions by System 2 are crucial to support System 1. Code for Reproduction: https://github.com/TsinghuaC3I/FS-GEN

comment: update figures and results on Pythia Series

♻ ☆ LocoMotion: Learning Motion-Focused Video-Language Representations ACCV 2024

Hazel Doughty, Fida Mohammad Thoker, Cees G. M. Snoek

This paper strives for motion-focused video-language representations. Existing methods to learn video-language representations use spatial-focused data, where identifying the objects and scene is often enough to distinguish the relevant caption. We instead propose LocoMotion to learn from motion-focused captions that describe the movement and temporal progression of local object motions. We achieve this by adding synthetic motions to videos and using the parameters of these motions to generate corresponding captions. Furthermore, we propose verb-variation paraphrasing to increase the caption variety and learn the link between primitive motions and high-level verbs. With this, we are able to learn a motion-focused video-language representation. Experiments demonstrate our approach is effective for a variety of downstream tasks, particularly when limited data is available for fine-tuning. Code is available: https://hazeldoughty.github.io/Papers/LocoMotion/

comment: ACCV 2024 Oral

♻ ☆ Reconfidencing LLMs from the Grouping Loss Perspective EMNLP 2024

Lihu Chen, Alexandre Perez-Lebel, Fabian M. Suchanek, Gaël Varoquaux

Large Language Models (LLMs), including ChatGPT and LLaMA, are susceptible to generating hallucinated answers in a confident tone. While efforts to elicit and calibrate confidence scores have proven useful, recent findings show that controlling uncertainty must go beyond calibration: predicted scores may deviate significantly from the actual posterior probabilities due to the impact of grouping loss. In this work, we construct a new evaluation dataset derived from a knowledge base to assess confidence scores given to answers of Mistral and LLaMA. Experiments show that they tend to be overconfident. Further, we show that they are more overconfident on some answers than others, \emph{eg} depending on the nationality of the person in the query. In uncertainty-quantification theory, this is grouping loss. To address this, we propose a solution to reconfidence LLMs, canceling not only calibration but also grouping loss. The LLMs, after the reconfidencing process, indicate improved confidence alignment with the accuracy of their responses.

comment: EMNLP 2024 Findings

♻ ☆ TravelPlanner: A Benchmark for Real-World Planning with Language Agents ICML 2024

Jian Xie, Kai Zhang, Jiangjie Chen, Tinghui Zhu, Renze Lou, Yuandong Tian, Yanghua Xiao, Yu Su

Planning has been part of the core pursuit for artificial intelligence since its conception, but earlier AI agents mostly focused on constrained settings because many of the cognitive substrates necessary for human-level planning have been lacking. Recently, language agents powered by large language models (LLMs) have shown interesting capabilities such as tool use and reasoning. Are these language agents capable of planning in more complex settings that are out of the reach of prior AI agents? To advance this investigation, we propose TravelPlanner, a new planning benchmark that focuses on travel planning, a common real-world planning scenario. It provides a rich sandbox environment, various tools for accessing nearly four million data records, and 1,225 meticulously curated planning intents and reference plans. Comprehensive evaluations show that the current language agents are not yet capable of handling such complex planning tasks-even GPT-4 only achieves a success rate of 0.6%. Language agents struggle to stay on task, use the right tools to collect information, or keep track of multiple constraints. However, we note that the mere possibility for language agents to tackle such a complex problem is in itself non-trivial progress. TravelPlanner provides a challenging yet meaningful testbed for future language agents.

comment: ICML 2024 (Spotlight)

♻ ☆ Trends in Integration of Knowledge and Large Language Models: A Survey and Taxonomy of Methods, Benchmarks, and Applications

Zhangyin Feng, Weitao Ma, Weijiang Yu, Lei Huang, Haotian Wang, Qianglong Chen, Weihua Peng, Xiaocheng Feng, Bing Qin, Ting liu

Large language models (LLMs) exhibit superior performance on various natural language tasks, but they are susceptible to issues stemming from outdated data and domain-specific limitations. In order to address these challenges, researchers have pursued two primary strategies, knowledge editing and retrieval augmentation, to enhance LLMs by incorporating external information from different aspects. Nevertheless, there is still a notable absence of a comprehensive survey. In this paper, we propose a review to discuss the trends in integration of knowledge and large language models, including taxonomy of methods, benchmarks, and applications. In addition, we conduct an in-depth analysis of different methods and point out potential research directions in the future. We hope this survey offers the community quick access and a comprehensive overview of this research area, with the intention of inspiring future research endeavors.

comment: Work in progress; 22 pages. This work has been submitted to the IEEE for possible publication

♻ ☆ Task Prompt Vectors: Effective Initialization through Multi-Task Soft-Prompt Transfer

Robert Belanec, Simon Ostermann, Ivan Srba, Maria Bielikova

Prompt tuning is an efficient solution for training large language models (LLMs). However, current soft-prompt-based methods often sacrifice multi-task modularity, requiring the training process to be fully or partially repeated for each newly added task. While recent work on task vectors applied arithmetic operations on full model weights to achieve the desired multi-task performance, a similar approach for soft-prompts is still missing. To this end, we introduce Task Prompt Vectors, created by element-wise difference between weights of tuned soft-prompts and their random initialization. Experimental results on 12 NLU datasets show that task prompt vectors can be used in low-resource settings to effectively initialize prompt tuning on similar tasks. In addition, we show that task prompt vectors are independent of the random initialization of prompt tuning on 2 different language model architectures. This allows prompt arithmetics with the pre-trained vectors from different tasks. In this way, we provide a competitive alternative to state-of-the-art baselines by arithmetic addition of task prompt vectors from multiple tasks.

♻ ☆ Let Me Teach You: Pedagogical Foundations of Feedback for Language Models EMNLP 2024

Beatriz Borges, Niket Tandon, Tanja Käser, Antoine Bosselut

Natural Language Feedback (NLF) is an increasingly popular mechanism for aligning Large Language Models (LLMs) to human preferences. Despite the diversity of the information it can convey, NLF methods are often hand-designed and arbitrary, with little systematic grounding. At the same time, research in learning sciences has long established several effective feedback models. In this opinion piece, we compile ideas from pedagogy to introduce FELT, a feedback framework for LLMs that outlines various characteristics of the feedback space, and a feedback content taxonomy based on these variables, providing a general mapping of the feedback space. In addition to streamlining NLF designs, FELT also brings out new, unexplored directions for research in NLF. We make our taxonomy available to the community, providing guides and examples for mapping our categorizations to future research.

comment: EMNLP 2024; 9 pages, 3 figures

♻ ☆ CPE-Pro: A Structure-Sensitive Deep Learning Method for Protein Representation and Origin Evaluation

Wenrui Gou, Wenhui Ge, Yang Tan, Mingchen Li, Guisheng Fan, Huiqun Yu

Protein structures are important for understanding their functions and interactions. Currently, many protein structure prediction methods are enriching the structure database. Discriminating the origin of structures is crucial for distinguishing between experimentally resolved and computationally predicted structures, evaluating the reliability of prediction methods, and guiding downstream biological studies. Building on works in structure prediction, We developed a structure-sensitive supervised deep learning model, Crystal vs Predicted Evaluator for Protein Structure (CPE-Pro), to represent and discriminate the origin of protein structures. CPE-Pro learns the structural information of proteins and captures inter-structural differences to achieve accurate traceability on four data classes, and is expected to be extended to more. Simultaneously, we utilized Foldseek to encode protein structures into "structure-sequences" and trained a protein Structural Sequence Language Model, SSLM. Preliminary experiments demonstrated that, compared to large-scale protein language models pre-trained on vast amounts of amino acid sequences, the "structure-sequence" enables the language model to learn more informative protein features, enhancing and optimizing structural representations. We have provided the code, model weights, and all related materials on https://github.com/GouWenrui/CPE-Pro-main.git.

♻ ☆ Do LLMs Have Distinct and Consistent Personality? TRAIT: Personality Testset designed for LLMs with Psychometrics

Seungbeen Lee, Seungwon Lim, Seungju Han, Giyeong Oh, Hyungjoo Chae, Jiwan Chung, Minju Kim, Beong-woo Kwak, Yeonsoo Lee, Dongha Lee, Jinyoung Yeo, Youngjae Yu

Recent advancements in Large Language Models (LLMs) have led to their adaptation in various domains as conversational agents. We wonder: can personality tests be applied to these agents to analyze their behavior, similar to humans? We introduce TRAIT, a new benchmark consisting of 8K multi-choice questions designed to assess the personality of LLMs. TRAIT is built on two psychometrically validated small human questionnaires, Big Five Inventory (BFI) and Short Dark Triad (SD-3), enhanced with the ATOMIC-10X knowledge graph to a variety of real-world scenarios. TRAIT also outperforms existing personality tests for LLMs in terms of reliability and validity, achieving the highest scores across four key metrics: Content Validity, Internal Validity, Refusal Rate, and Reliability. Using TRAIT, we reveal two notable insights into personalities of LLMs: 1) LLMs exhibit distinct and consistent personality, which is highly influenced by their training data (e.g., data used for alignment tuning), and 2) current prompting techniques have limited effectiveness in eliciting certain traits, such as high psychopathy or low conscientiousness, suggesting the need for further research in this direction.

comment: Preprint; Under review

♻ ☆ Attribute or Abstain: Large Language Models as Long Document Assistants EMNLP 2024

Jan Buchmann, Xiao Liu, Iryna Gurevych

LLMs can help humans working with long documents, but are known to hallucinate. Attribution can increase trust in LLM responses: The LLM provides evidence that supports its response, which enhances verifiability. Existing approaches to attribution have only been evaluated in RAG settings, where the initial retrieval confounds LLM performance. This is crucially different from the long document setting, where retrieval is not needed, but could help. Thus, a long document specific evaluation of attribution is missing. To fill this gap, we present LAB, a benchmark of 6 diverse long document tasks with attribution, and experiments with different approaches to attribution on 5 LLMs of different sizes. We find that citation, i.e. response generation and evidence extraction in one step, performs best for large and fine-tuned models, while additional retrieval can help for small, prompted models. We investigate whether the "Lost in the Middle'' phenomenon exists for attribution, but do not find this. We also find that evidence quality can predict response quality on datasets with simple responses, but not so for complex responses, as models struggle with providing evidence for complex claims.

comment: Accepted at EMNLP 2024. Code and data: https://github.com/UKPLab/arxiv2024-attribute-or-abstain

♻ ☆ I've Got 99 Problems But FLOPS Ain't One

Alexandru M. Gherghescu, Vlad-Andrei Bădoiu, Alexandru Agache, Mihai-Valentin Dumitru, Iuliu Vasilescu, Radu Mantu, Costin Raiciu

Hyperscalers dominate the landscape of large network deployments, yet they rarely share data or insights about the challenges they face. In light of this supremacy, what problems can we find to solve in this space? We take an unconventional approach to find relevant research directions, starting from public plans to build a $100 billion datacenter for machine learning applications. Leveraging the language models scaling laws, we discover what workloads such a datacenter might carry and explore the challenges one may encounter in doing so, with a focus on networking research. We conclude that building the datacenter and training such models is technically possible, but this requires novel wide-area transports for inter-DC communication, a multipath transport and novel datacenter topologies for intra-datacenter communication, high speed scale-up networks and transports, outlining a rich research agenda for the networking community.

♻ ☆ Interpreting Context Look-ups in Transformers: Investigating Attention-MLP Interactions EMNLP 2024

Clement Neo, Shay B. Cohen, Fazl Barez

Understanding the inner workings of large language models (LLMs) is crucial for advancing their theoretical foundations and real-world applications. While the attention mechanism and multi-layer perceptrons (MLPs) have been studied independently, their interactions remain largely unexplored. This study investigates how attention heads and next-token neurons interact in LLMs to predict new words. We propose a methodology to identify next-token neurons, find prompts that highly activate them, and determine the upstream attention heads responsible. We then generate and evaluate explanations for the activity of these attention heads in an automated manner. Our findings reveal that some attention heads recognize specific contexts relevant to predicting a token and activate a downstream token-predicting neuron accordingly. This mechanism provides a deeper understanding of how attention heads work with MLP neurons to perform next-token prediction. Our approach offers a foundation for further research into the intricate workings of LLMs and their impact on text generation and understanding.

comment: Accepted to EMNLP 2024 Main Conference

♻ ☆ Few-Shot Adversarial Prompt Learning on Vision-Language Models NeurIPS 2024

Yiwei Zhou, Xiaobo Xia, Zhiwei Lin, Bo Han, Tongliang Liu

The vulnerability of deep neural networks to imperceptible adversarial perturbations has attracted widespread attention. Inspired by the success of vision-language foundation models, previous efforts achieved zero-shot adversarial robustness by aligning adversarial visual features with text supervision. However, in practice, they are still unsatisfactory due to several issues, including heavy adaptation cost, suboptimal text supervision, and uncontrolled natural generalization capacity. In this paper, to address these issues, we propose a few-shot adversarial prompt framework where adapting input sequences with limited data makes significant adversarial robustness improvement. Specifically, we achieve this by providing adversarially correlated text supervision that is end-to-end learned from adversarial examples. We also propose a novel training objective that enhances the consistency of multi-modal features while encourages differentiated uni-modal features between natural and adversarial examples. The proposed framework gives access to learn adversarial text supervision, which provides superior cross-modal adversarial alignment and matches state-of-the-art zero-shot adversarial robustness with only 1% training data. Code is available at: https://github.com/lionel-w2/FAP.

comment: NeurIPS 2024

♻ ☆ Do Large Language Models Have an English Accent? Evaluating and Improving the Naturalness of Multilingual LLMs

Yanzhu Guo, Simone Conia, Zelin Zhou, Min Li, Saloni Potdar, Henry Xiao

Current Large Language Models (LLMs) are predominantly designed with English as the primary language, and even the few that are multilingual tend to exhibit strong English-centric biases. Much like speakers who might produce awkward expressions when learning a second language, LLMs often generate unnatural outputs in non-English languages, reflecting English-centric patterns in both vocabulary and grammar. Despite the importance of this issue, the naturalness of multilingual LLM outputs has received limited attention. In this paper, we address this gap by introducing novel automatic corpus-level metrics to assess the lexical and syntactic naturalness of LLM outputs in a multilingual context. Using our new metrics, we evaluate state-of-the-art LLMs on a curated benchmark in French and Chinese, revealing a tendency towards English-influenced patterns. To mitigate this issue, we also propose a simple and effective alignment method to improve the naturalness of an LLM in a target language and domain, achieving consistent improvements in naturalness without compromising the performance on general-purpose benchmarks. Our work highlights the importance of developing multilingual metrics, resources and methods for the new wave of multilingual LLMs.

♻ ☆ RaTEScore: A Metric for Radiology Report Generation EMNLP 2024

Weike Zhao, Chaoyi Wu, Xiaoman Zhang, Ya Zhang, Yanfeng Wang, Weidi Xie

This paper introduces a novel, entity-aware metric, termed as Radiological Report (Text) Evaluation (RaTEScore), to assess the quality of medical reports generated by AI models. RaTEScore emphasizes crucial medical entities such as diagnostic outcomes and anatomical details, and is robust against complex medical synonyms and sensitive to negation expressions. Technically, we developed a comprehensive medical NER dataset, RaTE-NER, and trained an NER model specifically for this purpose. This model enables the decomposition of complex radiological reports into constituent medical entities. The metric itself is derived by comparing the similarity of entity embeddings, obtained from a language model, based on their types and relevance to clinical significance. Our evaluations demonstrate that RaTEScore aligns more closely with human preference than existing metrics, validated both on established public benchmarks and our newly proposed RaTE-Eval benchmark.

comment: EMNLP 2024

♻ ☆ Can Language Models Induce Grammatical Knowledge from Indirect Evidence? EMNLP 2024

Miyu Oba, Yohei Oseki, Akiyo Fukatsu, Akari Haga, Hiroki Ouchi, Taro Watanabe, Saku Sugawara

What kinds of and how much data is necessary for language models to induce grammatical knowledge to judge sentence acceptability? Recent language models still have much room for improvement in their data efficiency compared to humans. This paper investigates whether language models efficiently use indirect data (indirect evidence), from which they infer sentence acceptability. In contrast, humans use indirect evidence efficiently, which is considered one of the inductive biases contributing to efficient language acquisition. To explore this question, we introduce the Wug InDirect Evidence Test (WIDET), a dataset consisting of training instances inserted into the pre-training data and evaluation instances. We inject synthetic instances with newly coined wug words into pretraining data and explore the model's behavior on evaluation data that assesses grammatical acceptability regarding those words. We prepare the injected instances by varying their levels of indirectness and quantity. Our experiments surprisingly show that language models do not induce grammatical knowledge even after repeated exposure to instances with the same structure but differing only in lexical items from evaluation instances in certain language phenomena. Our findings suggest a potential direction for future research: developing models that use latent indirect evidence to induce grammatical knowledge.

comment: This paper is accepted at EMNLP 2024 Main

♻ ☆ A Review of Prominent Paradigms for LLM-Based Agents: Tool Use (Including RAG), Planning, and Feedback Learning

Xinzhe Li

Tool use, planning, and feedback learning are currently three prominent paradigms for developing Large Language Model (LLM)-based agents across various tasks. Although numerous frameworks have been devised for each paradigm, their intricate workflows and inconsistent taxonomy create challenges in understanding and reviewing the frameworks across different paradigms. This survey introduces a unified taxonomy to systematically review and discuss these frameworks. Specifically, 1) the taxonomy defines environments/tasks, common LLM-profiled roles or LMPRs (policy models, evaluators, and dynamic models), and universally applicable workflows found in prior work, and 2) it enables a comparison of key perspectives on the implementations of LMPRs and workflow designs across different agent paradigms and frameworks. 3) Finally, we identify three limitations in existing workflow designs and systematically discuss the future work.

comment: Under Review

♻ ☆ GRAMMAR: Grounded and Modular Methodology for Assessment of Closed-Domain Retrieval-Augmented Language Model

Xinzhe Li, Ming Liu, Shang Gao

Retrieval-Augmented Generation (RAG) systems are widely used across various industries for querying closed-domain and in-house knowledge bases. However, evaluating these systems presents significant challenges due to the private nature of closed-domain data and a scarcity of queries with verifiable ground truths. Moreover, there is a lack of analytical methods to diagnose problematic modules and identify types of failure, such as those caused by knowledge deficits or issues with robustness. To address these challenges, we introduce GRAMMAR (GRounded And Modular Methodology for Assessment of RAG), an evaluation framework comprising a grounded data generation process and an evaluation protocol that effectively pinpoints defective modules. Our validation experiments reveal that GRAMMAR provides a reliable approach for identifying vulnerable modules and supports hypothesis testing for textual form vulnerabilities. An open-source tool accompanying this framework is available in our GitHub repository (see https://github.com/xinzhel/grammar), allowing for easy reproduction of our results and enabling reliable and modular evaluation in closed-domain settings.

comment: Under Review

♻ ☆ From Keywords to Structured Summaries: Streamlining Scholarly Information Access ISWC 2024

Mahsa Shamsabadi, Jennifer D'Souza

This paper highlights the growing importance of information retrieval (IR) engines in the scientific community, addressing the inefficiency of traditional keyword-based search engines due to the rising volume of publications. The proposed solution involves structured records, underpinning advanced information technology (IT) tools, including visualization dashboards, to revolutionize how researchers access and filter articles, replacing the traditional text-heavy approach. This vision is exemplified through a proof of concept centered on the "reproductive number estimate of infectious diseases" research theme, using a fine-tuned large language model (LLM) to automate the creation of structured records to populate a backend database that now goes beyond keywords. The result is a next-generation information access system as an IR method accessible at https://orkg.org/usecases/r0-estimates.

comment: 8 pages, 3 figures | Accepted for publication as a poster paper at the International Semantic Web Conference (ISWC 2024)

♻ ☆ Learning to Plan for Retrieval-Augmented Large Language Models from Knowledge Graphs EMNLP2024

Junjie Wang, Mingyang Chen, Binbin Hu, Dan Yang, Ziqi Liu, Yue Shen, Peng Wei, Zhiqiang Zhang, Jinjie Gu, Jun Zhou, Jeff Z. Pan, Wen Zhang, Huajun Chen

Improving the performance of large language models (LLMs) in complex question-answering (QA) scenarios has always been a research focal point. Recent studies have attempted to enhance LLMs' performance by combining step-wise planning with external retrieval. While effective for advanced models like GPT-3.5, smaller LLMs face challenges in decomposing complex questions, necessitating supervised fine-tuning. Previous work has relied on manual annotation and knowledge distillation from teacher LLMs, which are time-consuming and not accurate enough. In this paper, we introduce a novel framework for enhancing LLMs' planning capabilities by using planning data derived from knowledge graphs (KGs). LLMs fine-tuned with this data have improved planning capabilities, better equipping them to handle complex QA tasks that involve retrieval. Evaluations on multiple datasets, including our newly proposed benchmark, highlight the effectiveness of our framework and the benefits of KG-derived planning data.

comment: EMNLP2024 Findings

♻ ☆ Uncertainty Estimation and Quantification for LLMs: A Simple Supervised Approach

Linyu Liu, Yu Pan, Xiaocheng Li, Guanting Chen

In this paper, we study the problem of uncertainty estimation and calibration for LLMs. We begin by formulating the uncertainty estimation problem, a relevant yet underexplored area in existing literature. We then propose a supervised approach that leverages labeled datasets to estimate the uncertainty in LLMs' responses. Based on the formulation, we illustrate the difference between the uncertainty estimation for LLMs and that for standard ML models and explain why the hidden neurons of the LLMs may contain uncertainty information. Our designed approach demonstrates the benefits of utilizing hidden activations to enhance uncertainty estimation across various tasks and shows robust transferability in out-of-distribution settings. We distinguish the uncertainty estimation task from the uncertainty calibration task and show that better uncertainty estimation leads to better calibration performance. Furthermore, our method is easy to implement and adaptable to different levels of model accessibility including black box, grey box, and white box.

comment: 29 pages, 14 figures

♻ ☆ Regularizing Hidden States Enables Learning Generalizable Reward Model for LLMs NeurIPS 2024

Rui Yang, Ruomeng Ding, Yong Lin, Huan Zhang, Tong Zhang

Reward models trained on human preference data have been proven to effectively align Large Language Models (LLMs) with human intent within the framework of reinforcement learning from human feedback (RLHF). However, current reward models have limited generalization capabilities to unseen prompts and responses, which can lead to an unexpected phenomenon known as reward over-optimization, resulting in a decline in actual performance due to excessive optimization of rewards. While previous research has advocated for constraining policy optimization, our study introduces a novel approach to enhance the reward model's generalization ability against distribution shifts by regularizing the hidden states. Specifically, we retain the base model's language model head and incorporate a suite of text-generation losses to preserve the hidden states' text-generation capabilities, while concurrently learning a reward head behind the same hidden states. Our experimental results demonstrate that the introduced regularization technique markedly improves the accuracy of learned reward models across a variety of out-of-distribution (OOD) tasks and effectively alleviates the over-optimization issue in RLHF, offering a more reliable and robust preference learning paradigm.

comment: NeurIPS 2024

♻ ☆ Mind's Eye of LLMs: Visualization-of-Thought Elicits Spatial Reasoning in Large Language Models NeurIPS 2024

Wenshan Wu, Shaoguang Mao, Yadong Zhang, Yan Xia, Li Dong, Lei Cui, Furu Wei

Large language models (LLMs) have exhibited impressive performance in language comprehension and various reasoning tasks. However, their abilities in spatial reasoning, a crucial aspect of human cognition, remain relatively unexplored. Human possess a remarkable ability to create mental images of unseen objects and actions through a process known as the Mind's Eye, enabling the imagination of the unseen world. Inspired by this cognitive capacity, we propose Visualization-of-Thought (VoT) prompting. VoT aims to elicit spatial reasoning of LLMs by visualizing their reasoning traces, thereby guiding subsequent reasoning steps. We employed VoT for multi-hop spatial reasoning tasks, including natural language navigation, visual navigation, and visual tiling in 2D grid worlds. Experimental results demonstrated that VoT significantly enhances the spatial reasoning abilities of LLMs. Notably, VoT outperformed existing multimodal large language models (MLLMs) in these tasks. While VoT works surprisingly well on LLMs, the ability to generate mental images to facilitate spatial reasoning resembles the mind's eye process, suggesting its potential viability in MLLMs. Please find the dataset and codes at https://microsoft.github.io/visualization-of-thought

comment: 38th Conference on Neural Information Processing Systems (NeurIPS 2024)

♻ ☆ GPT-SW3: An Autoregressive Language Model for the Nordic Languages

Ariel Ekgren, Amaru Cuba Gyllensten, Felix Stollenwerk, Joey Öhman, Tim Isbister, Evangelia Gogoulou, Fredrik Carlsson, Alice Heiman, Judit Casademont, Magnus Sahlgren

This paper details the process of developing the first native large generative language model for the Nordic languages, GPT-SW3. We cover all parts of the development process, from data collection and processing, training configuration and instruction finetuning, to evaluation and considerations for release strategies. We hope that this paper can serve as a guide and reference for other researchers that undertake the development of large generative models for smaller languages.

♻ ☆ Generative AI Security: Challenges and Countermeasures

Banghua Zhu, Norman Mu, Jiantao Jiao, David Wagner

Generative AI's expanding footprint across numerous industries has led to both excitement and increased scrutiny. This paper delves into the unique security challenges posed by Generative AI, and outlines potential research directions for managing these risks.

♻ ☆ OpenMU: Your Swiss Army Knife for Music Understanding

Mengjie Zhao, Zhi Zhong, Zhuoyuan Mao, Shiqi Yang, Wei-Hsiang Liao, Shusuke Takahashi, Hiromi Wakaki, Yuki Mitsufuji

We present OpenMU-Bench, a large-scale benchmark suite for addressing the data scarcity issue in training multimodal language models to understand music. To construct OpenMU-Bench, we leveraged existing datasets and bootstrapped new annotations. OpenMU-Bench also broadens the scope of music understanding by including lyrics understanding and music tool usage. Using OpenMU-Bench, we trained our music understanding model, OpenMU, with extensive ablations, demonstrating that OpenMU outperforms baseline models such as MU-Llama. Both OpenMU and OpenMU-Bench are open-sourced to facilitate future research in music understanding and to enhance creative music production efficiency.

comment: Resources: https://github.com/mzhaojp22/openmu

♻ ☆ Reinforcement Learning with Dynamic Multi-Reward Weighting for Multi-Style Controllable Generation

Karin de Langis, Ryan Koo, Dongyeop Kang

Textual style expresses a diverse set of information, including interpersonal dynamics (e.g., formality) and the author's emotions or attitudes (e.g., disgust). An open question is how language models can be explicitly controlled so that they weave together target styles when generating text: for example, to produce text that is both negative and non-toxic. One approach to such controlled generation is multi-objective reinforcement learning (RL), but how best to combine multiple objectives in a reward function is an open question. In this paper, we investigate various formulations of multi-style rewards, including calibrated outputs from discriminators and dynamic weighting by discriminator gradient magnitudes. We find that our proposed dynamic weighting outperforms static weighting approaches with respect to style control while maintaining linguistic quality, and we explore its effectiveness in 2- and 3-style control.

♻ ☆ BrainTransformers: SNN-LLM

Zhengzheng Tang, Eva Zhu

This study introduces BrainTransformers, an innovative Large Language Model (LLM) implemented using Spiking Neural Networks (SNN). Our key contributions include: (1) designing SNN-compatible Transformer components such as SNNMatmul, SNNSoftmax, and SNNSiLU; (2) implementing an SNN approximation of the SiLU activation function; and (3) developing a Synapsis module to simulate synaptic plasticity. Our 3-billion parameter model, BrainTransformers-3B-Chat, demonstrates competitive performance across various benchmarks, including MMLU (63.2), BBH (54.1), ARC-C (54.3), and GSM8K (76.3), while potentially offering improved energy efficiency and biological plausibility. The model employs a three-stage training approach, including SNN-specific neuronal synaptic plasticity training. This research opens new avenues for brain-like AI systems in natural language processing and neuromorphic computing. Future work will focus on hardware optimization, developing specialized SNN fine-tuning tools, and exploring practical applications in energy-efficient computing environments.

♻ ☆ TSDS: Data Selection for Task-Specific Model Finetuning

Zifan Liu, Amin Karbasi, Theodoros Rekatsinas

Finetuning foundation models for specific tasks is an emerging paradigm in modern machine learning. The efficacy of task-specific finetuning largely depends on the selection of appropriate training data. We present TSDS (Task-Specific Data Selection), a framework to select data for task-specific model finetuning, guided by a small but representative set of examples from the target task. To do so, we formulate data selection for task-specific finetuning as an optimization problem with a distribution alignment loss based on optimal transport to capture the discrepancy between the selected data and the target distribution. In addition, we add a regularizer to encourage the diversity of the selected data and incorporate kernel density estimation into the regularizer to reduce the negative effects of near-duplicates among the candidate data. We connect our optimization problem to nearest neighbor search and design efficient algorithms to compute the optimal solution based on approximate nearest neighbor search techniques. We evaluate our method on data selection for both continued pretraining and instruction tuning of language models. We show that instruction tuning using data selected by our method with a 1% selection ratio often outperforms using the full dataset and beats the baseline selection methods by 1.5 points in F1 score on average.

comment: 31 pages, 1 figure

♻ ☆ Selective Vision is the Challenge for Visual Reasoning: A Benchmark for Visual Argument Understanding EMNLP 2024

Jiwan Chung, Sungjae Lee, Minseo Kim, Seungju Han, Ashkan Yousefpour, Jack Hessel, Youngjae Yu

Visual arguments, often used in advertising or social causes, rely on images to persuade viewers to do or believe something. Understanding these arguments requires selective vision: only specific visual stimuli within an image are relevant to the argument, and relevance can only be understood within the context of a broader argumentative structure. While visual arguments are readily appreciated by human audiences, we ask: are today's AI capable of similar understanding? We present VisArgs, a dataset of 1,611 images annotated with 5,112 visual premises (with regions), 5,574 commonsense premises, and reasoning trees connecting them into structured arguments. We propose three tasks for evaluating visual argument understanding: premise localization, premise identification, and conclusion deduction. Experiments show that 1) machines struggle to capture visual cues: GPT-4-O achieved 78.5% accuracy, while humans reached 98.0%. Models also performed 19.5% worse when distinguishing between irrelevant objects within the image compared to external objects. 2) Providing relevant visual premises improved model performance significantly.

comment: 12 pages, 6 figures. Accepted as main paper in EMNLP 2024

♻ ☆ Do LLMs Have Political Correctness? Analyzing Ethical Biases and Jailbreak Vulnerabilities in AI Systems

Isack Lee, Haebin Seong

Although large language models (LLMs) demonstrate impressive proficiency in various tasks, they present potential safety risks, such as `jailbreaks', where malicious inputs can coerce LLMs into generating harmful content. To address these issues, many LLM developers have implemented various safety measures to align these models. This alignment involves several techniques, including data filtering during pre-training, supervised fine-tuning, reinforcement learning from human feedback, and red-teaming exercises. These methods often introduce deliberate and intentional biases similar to Political Correctness (PC) to ensure the ethical behavior of LLMs. In this paper, we delve into the intentional biases injected into LLMs for safety purposes and examine methods to circumvent these safety alignment techniques. Notably, these intentional biases result in a jailbreaking success rate in GPT-4o models that differs by 20% between non-binary and cisgender keywords and by 16% between white and black keywords, even when the other parts of the prompts are identical. We introduce the concept of PCJailbreak, highlighting the inherent risks posed by these safety-induced biases. Additionally, we propose an efficient defense method PCDefense, which prevents jailbreak attempts by injecting defense prompts prior to generation. PCDefense stands as an appealing alternative to Guard Models, such as Llama-Guard, that require additional inference cost after text generation. Our findings emphasize the urgent need for LLM developers to adopt a more responsible approach when designing and implementing safety measures.

♻ ☆ When "Competency" in Reasoning Opens the Door to Vulnerability: Jailbreaking LLMs via Novel Complex Ciphers

Divij Handa, Zehua Zhang, Amir Saeidi, Chitta Baral

Recent advancements in the safety of Large Language Models (LLMs) have primarily focused on mitigating attacks crafted in natural language or in common encryption techniques like Base64. However, new models which often possess better reasoning capabilities, open the door to new attack vectors that were previously non-existent in older models. This seems counter-intuitive at first glance, but these advanced models can decipher more complex cryptic queries that previous models could not, making them susceptible to attacks using such prompts. To exploit this vulnerability, we propose Attacks using Custom Encryptions (ACE), a novel method to jailbreak LLMs by leveraging custom encryption schemes. We evaluate the effectiveness of ACE on four state-of-the-art LLMs, achieving Attack Success Rates (ASR) of up to 66% on close-source models and 88% on open-source models. Building upon this, we introduce Layered Attacks using Custom Encryptions (LACE), which employs multiple layers of encryption through our custom ciphers to further enhance the ASR. Our findings demonstrate that LACE significantly enhances the ability to jailbreak LLMs, increasing the ASR of GPT-4o from 40% to 78%, a 38% improvement. Our results highlight that the advanced capabilities of LLMs introduce unforeseen vulnerabilities to complex attacks. Specifically complex and layered ciphers increase the chance of jailbreaking.

comment: 14 pages, 7 figures

☆ DynamicCity: Large-Scale LiDAR Generation from Dynamic Scenes

Hengwei Bian, Lingdong Kong, Haozhe Xie, Liang Pan, Yu Qiao, Ziwei Liu

LiDAR scene generation has been developing rapidly recently. However, existing methods primarily focus on generating static and single-frame scenes, overlooking the inherently dynamic nature of real-world driving environments. In this work, we introduce DynamicCity, a novel 4D LiDAR generation framework capable of generating large-scale, high-quality LiDAR scenes that capture the temporal evolution of dynamic environments. DynamicCity mainly consists of two key models. 1) A VAE model for learning HexPlane as the compact 4D representation. Instead of using naive averaging operations, DynamicCity employs a novel Projection Module to effectively compress 4D LiDAR features into six 2D feature maps for HexPlane construction, which significantly enhances HexPlane fitting quality (up to 12.56 mIoU gain). Furthermore, we utilize an Expansion & Squeeze Strategy to reconstruct 3D feature volumes in parallel, which improves both network training efficiency and reconstruction accuracy than naively querying each 3D point (up to 7.05 mIoU gain, 2.06x training speedup, and 70.84% memory reduction). 2) A DiT-based diffusion model for HexPlane generation. To make HexPlane feasible for DiT generation, a Padded Rollout Operation is proposed to reorganize all six feature planes of the HexPlane as a squared 2D feature map. In particular, various conditions could be introduced in the diffusion or sampling process, supporting versatile 4D generation applications, such as trajectory- and command-driven generation, inpainting, and layout-conditioned generation. Extensive experiments on the CarlaSC and Waymo datasets demonstrate that DynamicCity significantly outperforms existing state-of-the-art 4D LiDAR generation methods across multiple metrics. The code will be released to facilitate future research.

comment: Preprint; 29 pages, 15 figures, 7 tables; Project Page at https://dynamic-city.github.io/

☆ FIPER: Generalizable Factorized Fields for Joint Image Compression and Super-Resolution

Yang-Che Sun, Cheng Yu Yeo, Ernie Chu, Jun-Cheng Chen, Yu-Lun Liu

In this work, we propose a unified representation for Super-Resolution (SR) and Image Compression, termed **Factorized Fields**, motivated by the shared principles between these two tasks. Both SISR and Image Compression require recovering and preserving fine image details--whether by enhancing resolution or reconstructing compressed data. Unlike previous methods that mainly focus on network architecture, our proposed approach utilizes a basis-coefficient decomposition to explicitly capture multi-scale visual features and structural components in images, addressing the core challenges of both tasks. We first derive our SR model, which includes a Coefficient Backbone and Basis Swin Transformer for generalizable Factorized Fields. Then, to further unify these two tasks, we leverage the strong information-recovery capabilities of the trained SR modules as priors in the compression pipeline, improving both compression efficiency and detail reconstruction. Additionally, we introduce a merged-basis compression branch that consolidates shared structures, further optimizing the compression process. Extensive experiments show that our unified representation delivers state-of-the-art performance, achieving an average relative improvement of 204.4% in PSNR over the baseline in Super-Resolution (SR) and 9.35% BD-rate reduction in Image Compression compared to the previous SOTA.

comment: Project page: https://jayisaking.github.io/FIPER/

☆ FreeVS: Generative View Synthesis on Free Driving Trajectory

Qitai Wang, Lue Fan, Yuqi Wang, Yuntao Chen, Zhaoxiang Zhang

Existing reconstruction-based novel view synthesis methods for driving scenes focus on synthesizing camera views along the recorded trajectory of the ego vehicle. Their image rendering performance will severely degrade on viewpoints falling out of the recorded trajectory, where camera rays are untrained. We propose FreeVS, a novel fully generative approach that can synthesize camera views on free new trajectories in real driving scenes. To control the generation results to be 3D consistent with the real scenes and accurate in viewpoint pose, we propose the pseudo-image representation of view priors to control the generation process. Viewpoint transformation simulation is applied on pseudo-images to simulate camera movement in each direction. Once trained, FreeVS can be applied to any validation sequences without reconstruction process and synthesis views on novel trajectories. Moreover, we propose two new challenging benchmarks tailored to driving scenes, which are novel camera synthesis and novel trajectory synthesis, emphasizing the freedom of viewpoints. Given that no ground truth images are available on novel trajectories, we also propose to evaluate the consistency of images synthesized on novel trajectories with 3D perception models. Experiments on the Waymo Open Dataset show that FreeVS has a strong image synthesis performance on both the recorded trajectories and novel trajectories. Project Page: https://freevs24.github.io/

comment: Project Page: https://freevs24.github.io/

☆ UnCLe: Unsupervised Continual Learning of Depth Completion

Suchisrit Gangopadhyay, Xien Chen, Michael Chu, Patrick Rim, Hyoungseob Park, Alex Wong

We propose UnCLe, a standardized benchmark for Unsupervised Continual Learning of a multimodal depth estimation task: Depth completion aims to infer a dense depth map from a pair of synchronized RGB image and sparse depth map. We benchmark depth completion models under the practical scenario of unsupervised learning over continuous streams of data. Existing methods are typically trained on a static, or stationary, dataset. However, when adapting to novel non-stationary distributions, they "catastrophically forget" previously learned information. UnCLe simulates these non-stationary distributions by adapting depth completion models to sequences of datasets containing diverse scenes captured from distinct domains using different visual and range sensors. We adopt representative methods from continual learning paradigms and translate them to enable unsupervised continual learning of depth completion. We benchmark these models for indoor and outdoor and investigate the degree of catastrophic forgetting through standard quantitative metrics. Furthermore, we introduce model inversion quality as an additional measure of forgetting. We find that unsupervised continual learning of depth completion is an open problem, and we invite researchers to leverage UnCLe as a development platform.

comment: Preprint

☆ WorldSimBench: Towards Video Generation Models as World Simulators

Yiran Qin, Zhelun Shi, Jiwen Yu, Xijun Wang, Enshen Zhou, Lijun Li, Zhenfei Yin, Xihui Liu, Lu Sheng, Jing Shao, Lei Bai, Wanli Ouyang, Ruimao Zhang

Recent advancements in predictive models have demonstrated exceptional capabilities in predicting the future state of objects and scenes. However, the lack of categorization based on inherent characteristics continues to hinder the progress of predictive model development. Additionally, existing benchmarks are unable to effectively evaluate higher-capability, highly embodied predictive models from an embodied perspective. In this work, we classify the functionalities of predictive models into a hierarchy and take the first step in evaluating World Simulators by proposing a dual evaluation framework called WorldSimBench. WorldSimBench includes Explicit Perceptual Evaluation and Implicit Manipulative Evaluation, encompassing human preference assessments from the visual perspective and action-level evaluations in embodied tasks, covering three representative embodied scenarios: Open-Ended Embodied Environment, Autonomous, Driving, and Robot Manipulation. In the Explicit Perceptual Evaluation, we introduce the HF-Embodied Dataset, a video assessment dataset based on fine-grained human feedback, which we use to train a Human Preference Evaluator that aligns with human perception and explicitly assesses the visual fidelity of World Simulators. In the Implicit Manipulative Evaluation, we assess the video-action consistency of World Simulators by evaluating whether the generated situation-aware video can be accurately translated into the correct control signals in dynamic environments. Our comprehensive evaluation offers key insights that can drive further innovation in video generation models, positioning World Simulators as a pivotal advancement toward embodied artificial intelligence.

☆ TP-Eval: Tap Multimodal LLMs' Potential in Evaluation by Customizing Prompts

Yuxuan Xie, Tianhua Li, Wenqi Shao, Kaipeng Zhang

Recently, multimodal large language models (MLLMs) have received much attention for their impressive capabilities. The evaluation of MLLMs is becoming critical to analyzing attributes of MLLMs and providing valuable insights. However, current benchmarks overlook the problem of prompt sensitivity - minor prompt variations may lead to significant performance fluctuations. Thus, inappropriate prompts may obscure the models' capabilities, underestimating the models' performance. Moreover, different models have different preferences for different prompts, and thus, using the same prompt for all models will cause evaluation bias. This paper analyzes this deficiency in existing benchmarks and further introduces a new evaluation framework named TP-Eval, which introduces a prompt customization method to reduce evaluation biases and tap models' potential. TP-Eval will rewrite the original prompts to different customized prompts for different models. In particular, we propose some well-designed modules for prompt customization tailored to the scenario of MLLM evaluation. Extensive experiments demonstrate the effectiveness of our approach to uncovering models' capabilities, and TP-Eval should benefit the community in developing more comprehensive and convincing MLLM evaluation benchmarks.

☆ SPIRE: Synergistic Planning, Imitation, and Reinforcement Learning for Long-Horizon Manipulation

Zihan Zhou, Animesh Garg, Dieter Fox, Caelan Garrett, Ajay Mandlekar

Robot learning has proven to be a general and effective technique for programming manipulators. Imitation learning is able to teach robots solely from human demonstrations but is bottlenecked by the capabilities of the demonstrations. Reinforcement learning uses exploration to discover better behaviors; however, the space of possible improvements can be too large to start from scratch. And for both techniques, the learning difficulty increases proportional to the length of the manipulation task. Accounting for this, we propose SPIRE, a system that first uses Task and Motion Planning (TAMP) to decompose tasks into smaller learning subproblems and second combines imitation and reinforcement learning to maximize their strengths. We develop novel strategies to train learning agents when deployed in the context of a planning system. We evaluate SPIRE on a suite of long-horizon and contact-rich robot manipulation problems. We find that SPIRE outperforms prior approaches that integrate imitation learning, reinforcement learning, and planning by 35% to 50% in average task performance, is 6 times more data efficient in the number of human demonstrations needed to train proficient agents, and learns to complete tasks nearly twice as efficiently. View https://sites.google.com/view/spire-corl-2024 for more details.

comment: Conference on Robot Learning (CoRL) 2024

☆ CLEAR: Character Unlearning in Textual and Visual Modalities

Alexey Dontsov, Dmitrii Korzh, Alexey Zhavoronkin, Boris Mikheev, Denis Bobkov, Aibek Alanov, Oleg Y. Rogov, Ivan Oseledets, Elena Tutubalina

Machine Unlearning (MU) is critical for enhancing privacy and security in deep learning models, particularly in large multimodal language models (MLLMs), by removing specific private or hazardous information. While MU has made significant progress in textual and visual modalities, multimodal unlearning (MMU) remains significantly underexplored, partially due to the absence of a suitable open-source benchmark. To address this, we introduce CLEAR, a new benchmark designed to evaluate MMU methods. CLEAR contains 200 fictitious individuals and 3,700 images linked with corresponding question-answer pairs, enabling a thorough evaluation across modalities. We assess 10 MU methods, adapting them for MMU, and highlight new challenges specific to multimodal forgetting. We also demonstrate that simple $\ell_1$ regularization on LoRA weights significantly mitigates catastrophic forgetting, preserving model performance on retained data. The dataset is available at https://huggingface.co/datasets/therem/CLEAR

☆ In-Pixel Foreground and Contrast Enhancement Circuits with Customizable Mapping

Md Rahatul Islam Udoy, Md Mazharul Islam, Elijah Johnson, Ahmedullah Aziz

This paper presents an innovative in-pixel contrast enhancement circuit that performs image processing directly within the pixel circuit. The circuit can be tuned for different modes of operation. In foreground enhancement mode, it suppresses low-intensity background pixels to nearly zero, isolating the foreground for better object visibility. In contrast enhancement mode, it improves overall image contrast. The contrast enhancement function is customizable both during the design phase and in real-time, allowing the circuit to adapt to specific applications and varying lighting conditions. A model of the designed pixel circuit is developed and applied to a full pixel array, demonstrating significant improvements in image quality. Simulations performed in HSPICE show a nearly 6x increase in Michelson Contrast Ratio (CR) in the foreground enhancement mode. The simulation results indicate its potential for real-time, adaptive contrast enhancement across various imaging environments.

☆ Real time anomalies detection on video

Fabien Poirier

Nowadays, many places use security cameras. Unfortunately, when an incident occurs, these technologies are used to show past events. So it can be considered as a deterrence tool than a detection tool. In this article, we will propose a deep learning approach trying to solve this problematic. This approach uses convolutional models (CNN) to extract relevant characteristics linked to the video images, theses characteristics will form times series to be analyzed by LSTM / GRU models.

☆ Scalable Ranked Preference Optimization for Text-to-Image Generation

Shyamgopal Karthik, Huseyin Coskun, Zeynep Akata, Sergey Tulyakov, Jian Ren, Anil Kag

Direct Preference Optimization (DPO) has emerged as a powerful approach to align text-to-image (T2I) models with human feedback. Unfortunately, successful application of DPO to T2I models requires a huge amount of resources to collect and label large-scale datasets, e.g., millions of generated paired images annotated with human preferences. In addition, these human preference datasets can get outdated quickly as the rapid improvements of T2I models lead to higher quality images. In this work, we investigate a scalable approach for collecting large-scale and fully synthetic datasets for DPO training. Specifically, the preferences for paired images are generated using a pre-trained reward function, eliminating the need for involving humans in the annotation process, greatly improving the dataset collection efficiency. Moreover, we demonstrate that such datasets allow averaging predictions across multiple models and collecting ranked preferences as opposed to pairwise preferences. Furthermore, we introduce RankDPO to enhance DPO-based methods using the ranking feedback. Applying RankDPO on SDXL and SD3-Medium models with our synthetically generated preference dataset ``Syn-Pic'' improves both prompt-following (on benchmarks like T2I-Compbench, GenEval, and DPG-Bench) and visual quality (through user studies). This pipeline presents a practical and scalable solution to develop better preference datasets to enhance the performance of text-to-image models.

comment: Project Page: https://snap-research.github.io/RankDPO/

☆ Characterization of the multiplicity of solutions for camera pose given two vertically-aligned landmarks and accelerometer

Alexander R. Pruss

We consider the problem of recovering the position and orientation of a camera equipped with an accelerometer from sensor images of two labeled landmarks whose positions in a coordinate system aligned in a known way with gravity are known. This a variant on the much studied P$n$P problem of recovering camera position and orientation from $n$ points without any gravitational data. It is proved that in three types of singular cases there are infinitely many solutions, in another type of case there is one, and in a final type of case there are two. A precise characterization of each type of case. In particular, there is always a unique solution in the practically interesting case where the two landmarks are at the same altitude and the camera is at a different altitude. This case is studied by numerical simulation and an implementation on a consumer cellphone. It is also proved that if the two landmarks are unlabeled, then apart from the same singular cases, there are still always one or two solutions.

comment: 32 pages, 8 figures

☆ A Pipeline for Segmenting and Structuring RGB-D Data for Robotics Applications

Zhiwu Zheng, Lauren Mentzer, Berk Iskender, Michael Price, Colm Prendergast, Audren Cloitre

We introduce a novel pipeline for segmenting and structuring color and depth (RGB-D) data. Existing processing pipelines for RGB-D data have focused on extracting geometric information alone. This approach precludes the development of more advanced robotic navigation and manipulation algorithms, which benefit from a semantic understanding of their environment. Our pipeline can segment RGB-D data into accurate semantic masks. These masks are then used to fuse raw captured point clouds into semantically separated point clouds. We store this information using the Universal Scene Description (USD) file format, a format suitable for easy querying by downstream robotics algorithms, human-friendly visualization, and robotics simulation.

☆ Robust Two-View Geometry Estimation with Implicit Differentiation IROS 2024

Vladislav Pyatov, Iaroslav Koshelev, Stamatis Lefkimmiatis

We present a novel two-view geometry estimation framework which is based on a differentiable robust loss function fitting. We propose to treat the robust fundamental matrix estimation as an implicit layer, which allows us to avoid backpropagation through time and significantly improves the numerical stability. To take full advantage of the information from the feature matching stage we incorporate learnable weights that depend on the matching confidences. In this way our solution brings together feature extraction, matching and two-view geometry estimation in a unified end-to-end trainable pipeline. We evaluate our approach on the camera pose estimation task in both outdoor and indoor scenarios. The experiments on several datasets show that the proposed method outperforms both classic and learning-based state-of-the-art methods by a large margin. The project webpage is available at: https://github.com/VladPyatov/ihls

comment: IROS 2024 Accepted

☆ A Wavelet Diffusion GAN for Image Super-Resolution

Lorenzo Aloisi, Luigi Sigillo, Aurelio Uncini, Danilo Comminiello

In recent years, diffusion models have emerged as a superior alternative to generative adversarial networks (GANs) for high-fidelity image generation, with wide applications in text-to-image generation, image-to-image translation, and super-resolution. However, their real-time feasibility is hindered by slow training and inference speeds. This study addresses this challenge by proposing a wavelet-based conditional Diffusion GAN scheme for Single-Image Super-Resolution (SISR). Our approach utilizes the diffusion GAN paradigm to reduce the timesteps required by the reverse diffusion process and the Discrete Wavelet Transform (DWT) to achieve dimensionality reduction, decreasing training and inference times significantly. The results of an experimental validation on the CelebA-HQ dataset confirm the effectiveness of our proposed scheme. Our approach outperforms other state-of-the-art methodologies successfully ensuring high-fidelity output while overcoming inherent drawbacks associated with diffusion models in time-sensitive applications.

comment: The paper has been accepted at Italian Workshop on Neural Networks (WIRN) 2024

☆ Medical Imaging Complexity and its Effects on GAN Performance ACCV

William Cagas, Chan Ko, Blake Hsiao, Shryuk Grandhi, Rishi Bhattacharya, Kevin Zhu, Michael Lam

The proliferation of machine learning models in diverse clinical applications has led to a growing need for high-fidelity, medical image training data. Such data is often scarce due to cost constraints and privacy concerns. Alleviating this burden, medical image synthesis via generative adversarial networks (GANs) emerged as a powerful method for synthetically generating photo-realistic images based on existing sets of real medical images. However, the exact image set size required to efficiently train such a GAN is unclear. In this work, we experimentally establish benchmarks that measure the relationship between a sample dataset size and the fidelity of the generated images, given the dataset's distribution of image complexities. We analyze statistical metrics based on delentropy, an image complexity measure rooted in Shannon's entropy in information theory. For our pipeline, we conduct experiments with two state-of-the-art GANs, StyleGAN 3 and SPADE-GAN, trained on multiple medical imaging datasets with variable sample sizes. Across both GANs, general performance improved with increasing training set size but suffered with increasing complexity.

comment: Accepted to ACCV, Workshop on Generative AI for Synthetic Medical Data

☆ VR-Splatting: Foveated Radiance Field Rendering via 3D Gaussian Splatting and Neural Points

Linus Franke, Laura Fink, Marc Stamminger

Recent advances in novel view synthesis (NVS), particularly neural radiance fields (NeRF) and Gaussian splatting (3DGS), have demonstrated impressive results in photorealistic scene rendering. These techniques hold great potential for applications in virtual tourism and teleportation, where immersive realism is crucial. However, the high-performance demands of virtual reality (VR) systems present challenges in directly utilizing even such fast-to-render scene representations like 3DGS due to latency and computational constraints. In this paper, we propose foveated rendering as a promising solution to these obstacles. We analyze state-of-the-art NVS methods with respect to their rendering performance and compatibility with the human visual system. Our approach introduces a novel foveated rendering approach for Virtual Reality, that leverages the sharp, detailed output of neural point rendering for the foveal region, fused with a smooth rendering of 3DGS for the peripheral vision. Our evaluation confirms that perceived sharpness and detail-richness are increased by our approach compared to a standard VR-ready 3DGS configuration. Our system meets the necessary performance requirements for real-time VR interactions, ultimately enhancing the user's immersive experience. Project page: https://lfranke.github.io/vr_splatting

☆ Gaze-Assisted Medical Image Segmentation NeurIPS'24

Leila Khaertdinova, Ilya Pershin, Tatiana Shmykova, Bulat Ibragimov

The annotation of patient organs is a crucial part of various diagnostic and treatment procedures, such as radiotherapy planning. Manual annotation is extremely time-consuming, while its automation using modern image analysis techniques has not yet reached levels sufficient for clinical adoption. This paper investigates the idea of semi-supervised medical image segmentation using human gaze as interactive input for segmentation correction. In particular, we fine-tuned the Segment Anything Model in Medical Images (MedSAM), a public solution that uses various prompt types as additional input for semi-automated segmentation correction. We used human gaze data from reading abdominal images as a prompt for fine-tuning MedSAM. The model was validated on a public WORD database, which consists of 120 CT scans of 16 abdominal organs. The results of the gaze-assisted MedSAM were shown to be superior to the results of the state-of-the-art segmentation models. In particular, the average Dice coefficient for 16 abdominal organs was 85.8%, 86.7%, 81.7%, and 90.5% for nnUNetV2, ResUNet, original MedSAM, and our gaze-assisted MedSAM model, respectively.

comment: 16 pages, 4 figures, Accepted to AIM-FM Workshop @ NeurIPS'24

☆ Addressing Asynchronicity in Clinical Multimodal Fusion via Individualized Chest X-ray Generation NeurIPS-24

Wenfang Yao, Chen Liu, Kejing Yin, William K. Cheung, Jing Qin

Integrating multi-modal clinical data, such as electronic health records (EHR) and chest X-ray images (CXR), is particularly beneficial for clinical prediction tasks. However, in a temporal setting, multi-modal data are often inherently asynchronous. EHR can be continuously collected but CXR is generally taken with a much longer interval due to its high cost and radiation dose. When clinical prediction is needed, the last available CXR image might have been outdated, leading to suboptimal predictions. To address this challenge, we propose DDL-CXR, a method that dynamically generates an up-to-date latent representation of the individualized CXR images. Our approach leverages latent diffusion models for patient-specific generation strategically conditioned on a previous CXR image and EHR time series, providing information regarding anatomical structures and disease progressions, respectively. In this way, the interaction across modalities could be better captured by the latent CXR generation process, ultimately improving the prediction performance. Experiments using MIMIC datasets show that the proposed model could effectively address asynchronicity in multimodal fusion and consistently outperform existing methods.

comment: Accepted by NeurIPS-24

☆ R-CoT: Reverse Chain-of-Thought Problem Generation for Geometric Reasoning in Large Multimodal Models

Linger Deng, Yuliang Liu, Bohan Li, Dongliang Luo, Liang Wu, Chengquan Zhang, Pengyuan Lyu, Ziyang Zhang, Gang Zhang, Errui Ding, Yingying Zhu, Xiang Bai

Existing Large Multimodal Models (LMMs) struggle with mathematical geometric reasoning due to a lack of high-quality image-text paired data. Current geometric data generation approaches, which apply preset templates to generate geometric data or use Large Language Models (LLMs) to rephrase questions and answers (Q&A), unavoidably limit data accuracy and diversity. To synthesize higher-quality data, we propose a two-stage Reverse Chain-of-Thought (R-CoT) geometry problem generation pipeline. First, we introduce GeoChain to produce high-fidelity geometric images and corresponding descriptions highlighting relations among geometric elements. We then design a Reverse A&Q method that reasons step-by-step based on the descriptions and generates questions in reverse from the reasoning results. Experiments demonstrate that the proposed method brings significant and consistent improvements on multiple LMM baselines, achieving new performance records in the 2B, 7B, and 8B settings. Notably, R-CoT-8B significantly outperforms previous state-of-the-art open-source mathematical models by 16.6% on MathVista and 9.2% on GeoQA, while also surpassing the closed-source model GPT-4o by an average of 13% across both datasets. The code is available at https://github.com/dle666/R-CoT.

☆ A utility-based spatial analysis of residential street-level conditions; A case study of Rotterdam

Sander van Cranenburgh, Francisco Garrido-Valenzuela

Residential location choices are traditionally modelled using factors related to accessibility and socioeconomic environments, neglecting the importance of local street-level conditions. Arguably, this neglect is due to data practices. Today, however, street-level images -- which are highly effective at encoding street-level conditions -- are widely available. Additionally, recent advances in discrete choice models incorporating computer vision capabilities offer opportunities to integrate street-level conditions into residential location choice analysis. This study leverages these developments to investigate the spatial distribution of utility derived from street-level conditions in residential location choices on a city-wide scale. In our case study of Rotterdam, the Netherlands, we find that the utility derived from street-level conditions varies significantly on a highly localised scale, with conditions rapidly changing even within neighbourhoods. Our results also reveal that the high real-estate prices in the city centre cannot be attributed to attractive street-level conditions. Furthermore, whereas the city centre is characterised by relatively unattractive residential street-level conditions, neighbourhoods in the southern part of the city -- often perceived as problematic -- exhibit surprisingly appealing street-level environments. The methodological contribution of this paper is that it advances the discrete choice models incorporating computer vision capabilities by introducing a semantic regularisation layer to the model. Thereby, it adds explainability and eliminates the need for a separate pipeline to extract information from images, streamlining the analysis. As such, this paper's findings and methodological advancements pave the way for further studies to explore integrating street-level conditions in urban planning.

☆ CASCRNet: An Atrous Spatial Pyramid Pooling and Shared Channel Residual based Network for Capsule Endoscopy

K V Srinanda, M Manvith Prabhu, Shyam Lal

This manuscript summarizes work on the Capsule Vision Challenge 2024 by MISAHUB. To address the multi-class disease classification task, which is challenging due to the complexity and imbalance in the Capsule Vision challenge dataset, this paper proposes CASCRNet (Capsule endoscopy-Aspp-SCR-Network), a parameter-efficient and novel model that uses Shared Channel Residual (SCR) blocks and Atrous Spatial Pyramid Pooling (ASPP) blocks. Further, the performance of the proposed model is compared with other well-known approaches. The experimental results yield that proposed model provides better disease classification results. The proposed model was successful in classifying diseases with an F1 Score of 78.5% and a Mean AUC of 98.3%, which is promising given its compact architecture.

comment: 8 pages, 4 figures

☆ Blendify -- Python rendering framework for Blender

Vladimir Guzov, Ilya A. Petrov, Gerard Pons-Moll

With the rapid growth of the volume of research fields like computer vision and computer graphics, researchers require effective and user-friendly rendering tools to visualize results. While advanced tools like Blender offer powerful capabilities, they also require a significant effort to master. This technical report introduces Blendify, a lightweight Python-based framework that seamlessly integrates with Blender, providing a high-level API for scene creation and rendering. Blendify reduces the complexity of working with Blender's native API by automating object creation, handling the colors and material linking, and implementing features such as shadow-catcher objects while maintaining support for high-quality ray-tracing rendering output. With a focus on usability Blendify enables efficient and flexible rendering workflow for rendering in common computer vision and computer graphics use cases. The code is available at https://github.com/ptrvilya/blendify

comment: Project page: https://virtualhumans.mpi-inf.mpg.de/blendify/

☆ ROCKET-1: Master Open-World Interaction with Visual-Temporal Context Prompting

Shaofei Cai, Zihao Wang, Kewei Lian, Zhancun Mu, Xiaojian Ma, Anji Liu, Yitao Liang

Vision-language models (VLMs) have excelled in multimodal tasks, but adapting them to embodied decision-making in open-world environments presents challenges. A key issue is the difficulty in smoothly connecting individual entities in low-level observations with abstract concepts required for planning. A common approach to address this problem is through the use of hierarchical agents, where VLMs serve as high-level reasoners that break down tasks into executable sub-tasks, typically specified using language and imagined observations. However, language often fails to effectively convey spatial information, while generating future images with sufficient accuracy remains challenging. To address these limitations, we propose visual-temporal context prompting, a novel communication protocol between VLMs and policy models. This protocol leverages object segmentation from both past and present observations to guide policy-environment interactions. Using this approach, we train ROCKET-1, a low-level policy that predicts actions based on concatenated visual observations and segmentation masks, with real-time object tracking provided by SAM-2. Our method unlocks the full potential of VLMs visual-language reasoning abilities, enabling them to solve complex creative tasks, especially those heavily reliant on spatial understanding. Experiments in Minecraft demonstrate that our approach allows agents to accomplish previously unattainable tasks, highlighting the effectiveness of visual-temporal context prompting in embodied decision-making. Codes and demos will be available on the project page: https://craftjarvis.github.io/ROCKET-1.

☆ TAGE: Trustworthy Attribute Group Editing for Stable Few-shot Image Generation

Ruicheng Zhang, Guoheng Huang, Yejing Huo, Xiaochen Yuan, Zhizhen Zhou, Xuhang Chen, Guo Zhong

Generative Adversarial Networks (GANs) have emerged as a prominent research focus for image editing tasks, leveraging the powerful image generation capabilities of the GAN framework to produce remarkable results.However, prevailing approaches are contingent upon extensive training datasets and explicit supervision, presenting a significant challenge in manipulating the diverse attributes of new image classes with limited sample availability. To surmount this hurdle, we introduce TAGE, an innovative image generation network comprising three integral modules: the Codebook Learning Module (CLM), the Code Prediction Module (CPM) and the Prompt-driven Semantic Module (PSM). The CPM module delves into the semantic dimensions of category-agnostic attributes, encapsulating them within a discrete codebook. This module is predicated on the concept that images are assemblages of attributes, and thus, by editing these category-independent attributes, it is theoretically possible to generate images from unseen categories. Subsequently, the CPM module facilitates naturalistic image editing by predicting indices of category-independent attribute vectors within the codebook. Additionally, the PSM module generates semantic cues that are seamlessly integrated into the Transformer architecture of the CPM, enhancing the model's comprehension of the targeted attributes for editing. With these semantic cues, the model can generate images that accentuate desired attributes more prominently while maintaining the integrity of the original category, even with a limited number of samples. We have conducted extensive experiments utilizing the Animal Faces, Flowers, and VGGFaces datasets. The results of these experiments demonstrate that our proposed method not only achieves superior performance but also exhibits a high degree of stability when compared to other few-shot image generation techniques.

comment: Accepted by International Conference on Signal Processing Systems Conference

☆ Few-shot NeRF by Adaptive Rendering Loss Regularization ECCV2024

Qingshan Xu, Xuanyu Yi, Jianyao Xu, Wenbing Tao, Yew-Soon Ong, Hanwang Zhang

Novel view synthesis with sparse inputs poses great challenges to Neural Radiance Field (NeRF). Recent works demonstrate that the frequency regularization of Positional Encoding (PE) can achieve promising results for few-shot NeRF. In this work, we reveal that there exists an inconsistency between the frequency regularization of PE and rendering loss. This prevents few-shot NeRF from synthesizing higher-quality novel views. To mitigate this inconsistency, we propose Adaptive Rendering loss regularization for few-shot NeRF, dubbed AR-NeRF. Specifically, we present a two-phase rendering supervision and an adaptive rendering loss weight learning strategy to align the frequency relationship between PE and 2D-pixel supervision. In this way, AR-NeRF can learn global structures better in the early training phase and adaptively learn local details throughout the training process. Extensive experiments show that our AR-NeRF achieves state-of-the-art performance on different datasets, including object-level and complex scenes.

comment: Accepted by ECCV2024

☆ Exploiting Text-Image Latent Spaces for the Description of Visual Concepts ICPR

Laines Schmalwasser, Jakob Gawlikowski, Joachim Denzler, Julia Niebling

Concept Activation Vectors (CAVs) offer insights into neural network decision-making by linking human friendly concepts to the model's internal feature extraction process. However, when a new set of CAVs is discovered, they must still be translated into a human understandable description. For image-based neural networks, this is typically done by visualizing the most relevant images of a CAV, while the determination of the concept is left to humans. In this work, we introduce an approach to aid the interpretation of newly discovered concept sets by suggesting textual descriptions for each CAV. This is done by mapping the most relevant images representing a CAV into a text-image embedding where a joint description of these relevant images can be computed. We propose utilizing the most relevant receptive fields instead of full images encoded. We demonstrate the capabilities of this approach in multiple experiments with and without given CAV labels, showing that the proposed approach provides accurate descriptions for the CAVs and reduces the challenge of concept interpretation.

comment: 19 pages, 7 figures, to be published in ICPR

☆ Att2CPC: Attention-Guided Lossy Attribute Compression of Point Clouds

Kai Liu, Kang You, Pan Gao, Manoranjan Paul

With the great progress of 3D sensing and acquisition technology, the volume of point cloud data has grown dramatically, which urges the development of efficient point cloud compression methods. In this paper, we focus on the task of learned lossy point cloud attribute compression (PCAC). We propose an efficient attention-based method for lossy compression of point cloud attributes leveraging on an autoencoder architecture. Specifically, at the encoding side, we conduct multiple downsampling to best exploit the local attribute patterns, in which effective External Cross Attention (ECA) is devised to hierarchically aggregate features by intergrating attributes and geometry contexts. At the decoding side, the attributes of the point cloud are progressively reconstructed based on the multi-scale representation and the zero-padding upsampling tactic. To the best of our knowledge, this is the first approach to introduce attention mechanism to point-based lossy PCAC task. We verify the compression efficiency of our model on various sequences, including human body frames, sparse objects, and large-scale point cloud scenes. Experiments show that our method achieves an average improvement of 1.15 dB and 2.13 dB in BD-PSNR of Y channel and YUV channel, respectively, when comparing with the state-of-the-art point-based method Deep-PCAC. Codes of this paper are available at https://github.com/I2-Multimedia-Lab/Att2CPC.

☆ DREB-Net: Dual-stream Restoration Embedding Blur-feature Fusion Network for High-mobility UAV Object Detection

Qingpeng Li, Yuxin Zhang, Leyuan Fang, Yuhan Kang, Shutao Li, Xiao Xiang Zhu

Object detection algorithms are pivotal components of unmanned aerial vehicle (UAV) imaging systems, extensively employed in complex fields. However, images captured by high-mobility UAVs often suffer from motion blur cases, which significantly impedes the performance of advanced object detection algorithms. To address these challenges, we propose an innovative object detection algorithm specifically designed for blurry images, named DREB-Net (Dual-stream Restoration Embedding Blur-feature Fusion Network). First, DREB-Net addresses the particularities of blurry image object detection problem by incorporating a Blurry image Restoration Auxiliary Branch (BRAB) during the training phase. Second, it fuses the extracted shallow features via Multi-level Attention-Guided Feature Fusion (MAGFF) module, to extract richer features. Here, the MAGFF module comprises local attention modules and global attention modules, which assign different weights to the branches. Then, during the inference phase, the deep feature extraction of the BRAB can be removed to reduce computational complexity and improve detection speed. In loss function, a combined loss of MSE and SSIM is added to the BRAB to restore blurry images. Finally, DREB-Net introduces Fast Fourier Transform in the early stages of feature extraction, via a Learnable Frequency domain Amplitude Modulation Module (LFAMM), to adjust feature amplitude and enhance feature processing capability. Experimental results indicate that DREB-Net can still effectively perform object detection tasks under motion blur in captured images, showcasing excellent performance and broad application prospects. Our source code will be available at https://github.com/EEIC-Lab/DREB-Net.git.

☆ Deep Learning for Active Region Classification: A Systematic Study from Convolutional Neural Networks to Vision Transformers

Edoardo Legnaro, Sabrina Guastavino, Michele Piana, Anna Maria Massone

A solar active region can significantly disrupt the Sun Earth space environment, often leading to severe space weather events such as solar flares and coronal mass ejections. As a consequence, the automatic classification of active region groups is the crucial starting point for accurately and promptly predicting solar activity. This study presents our results concerned with the application of deep learning techniques to the classification of active region cutouts based on the Mount Wilson classification scheme. Specifically, we have explored the latest advancements in image classification architectures, from Convolutional Neural Networks to Vision Transformers, and reported on their performances for the active region classification task, showing that the crucial point for their effectiveness consists in a robust training process based on the latest advances in the field.

☆ Learning Lossless Compression for High Bit-Depth Volumetric Medical Image

Kai Wang, Yuanchao Bai, Daxin Li, Deming Zhai, Junjun Jiang, Xianming Liu

Recent advances in learning-based methods have markedly enhanced the capabilities of image compression. However, these methods struggle with high bit-depth volumetric medical images, facing issues such as degraded performance, increased memory demand, and reduced processing speed. To address these challenges, this paper presents the Bit-Division based Lossless Volumetric Image Compression (BD-LVIC) framework, which is tailored for high bit-depth medical volume compression. The BD-LVIC framework skillfully divides the high bit-depth volume into two lower bit-depth segments: the Most Significant Bit-Volume (MSBV) and the Least Significant Bit-Volume (LSBV). The MSBV concentrates on the most significant bits of the volumetric medical image, capturing vital structural details in a compact manner. This reduction in complexity greatly improves compression efficiency using traditional codecs. Conversely, the LSBV deals with the least significant bits, which encapsulate intricate texture details. To compress this detailed information effectively, we introduce an effective learning-based compression model equipped with a Transformer-Based Feature Alignment Module, which exploits both intra-slice and inter-slice redundancies to accurately align features. Subsequently, a Parallel Autoregressive Coding Module merges these features to precisely estimate the probability distribution of the least significant bit-planes. Our extensive testing demonstrates that the BD-LVIC framework not only sets new performance benchmarks across various datasets but also maintains a competitive coding speed, highlighting its significant potential and practical utility in the realm of volumetric medical image compression.

comment: 13 pages

☆ PGDiffSeg: Prior-Guided Denoising Diffusion Model with Parameter-Shared Attention for Breast Cancer Segmentation

Feiyan Feng, Tianyu Liu, Hong Wang, Jun Zhao, Wei Li, Yanshen Sun

Early detection through imaging and accurate diagnosis is crucial in mitigating the high mortality rate associated with breast cancer. However, locating tumors from low-resolution and high-noise medical images is extremely challenging. Therefore, this paper proposes a novel PGDiffSeg (Prior-Guided Diffusion Denoising Model with Parameter-Shared Attention) that applies diffusion denoising methods to breast cancer medical image segmentation, accurately recovering the affected areas from Gaussian noise. Firstly, we design a parallel pipeline for noise processing and semantic information processing and propose a parameter-shared attention module (PSA) in multi-layer that seamlessly integrates these two pipelines. This integration empowers PGDiffSeg to incorporate semantic details at multiple levels during the denoising process, producing highly accurate segmentation maps. Secondly, we introduce a guided strategy that leverages prior knowledge to simulate the decision-making process of medical professionals, thereby enhancing the model's ability to locate tumor positions precisely. Finally, we provide the first-ever discussion on the interpretability of the generative diffusion model in the context of breast cancer segmentation. Extensive experiments have demonstrated the superiority of our model over the current state-of-the-art approaches, confirming its effectiveness as a flexible diffusion denoising method suitable for medical image research. Our code will be publicly available later.

☆ EntityCLIP: Entity-Centric Image-Text Matching via Multimodal Attentive Contrastive Learning

Yaxiong Wang, Yaxiong Wang, Lianwei Wu, Lechao Cheng, Zhun Zhong, Meng Wang

Recent advancements in image-text matching have been notable, yet prevailing models predominantly cater to broad queries and struggle with accommodating fine-grained query intention. In this paper, we work towards the \textbf{E}ntity-centric \textbf{I}mage-\textbf{T}ext \textbf{M}atching (EITM), a task that the text and image involve specific entity-related information. The challenge of this task mainly lies in the larger semantic gap in entity association modeling, comparing with the general image-text matching problem.To narrow the huge semantic gap between the entity-centric text and the images, we take the fundamental CLIP as the backbone and devise a multimodal attentive contrastive learning framework to tam CLIP to adapt EITM problem, developing a model named EntityCLIP. The key of our multimodal attentive contrastive learning is to generate interpretive explanation text using Large Language Models (LLMs) as the bridge clues. In specific, we proceed by extracting explanatory text from off-the-shelf LLMs. This explanation text, coupled with the image and text, is then input into our specially crafted Multimodal Attentive Experts (MMAE) module, which effectively integrates explanation texts to narrow the gap of the entity-related text and image in a shared semantic space. Building on the enriched features derived from MMAE, we further design an effective Gated Integrative Image-text Matching (GI-ITM) strategy. The GI-ITM employs an adaptive gating mechanism to aggregate MMAE's features, subsequently applying image-text matching constraints to steer the alignment between the text and the image. Extensive experiments are conducted on three social media news benchmarks including N24News, VisualNews, and GoodNews, the results shows that our method surpasses the competition methods with a clear margin.

☆ An Intelligent Agentic System for Complex Image Restoration Problems

Kaiwen Zhu, Jinjin Gu, Zhiyuan You, Yu Qiao, Chao Dong

Real-world image restoration (IR) is inherently complex and often requires combining multiple specialized models to address diverse degradations. Inspired by human problem-solving, we propose AgenticIR, an agentic system that mimics the human approach to image processing by following five key stages: Perception, Scheduling, Execution, Reflection, and Rescheduling. AgenticIR leverages large language models (LLMs) and vision-language models (VLMs) that interact via text generation to dynamically operate a toolbox of IR models. We fine-tune VLMs for image quality analysis and employ LLMs for reasoning, guiding the system step by step. To compensate for LLMs' lack of specific IR knowledge and experience, we introduce a self-exploration method, allowing the LLM to observe and summarize restoration results into referenceable documents. Experiments demonstrate AgenticIR's potential in handling complex IR tasks, representing a promising path toward achieving general intelligence in visual processing.

☆ GenUDC: High Quality 3D Mesh Generation with Unsigned Dual Contouring Representation

Ruowei Wang, Jiaqi Li, Dan Zeng, Xueqi Ma, Zixiang Xu, Jianwei Zhang, Qijun Zhao

Generating high-quality meshes with complex structures and realistic surfaces is the primary goal of 3D generative models. Existing methods typically employ sequence data or deformable tetrahedral grids for mesh generation. However, sequence-based methods have difficulty producing complex structures with many faces due to memory limits. The deformable tetrahedral grid-based method MeshDiffusion fails to recover realistic surfaces due to the inherent ambiguity in deformable grids. We propose the GenUDC framework to address these challenges by leveraging the Unsigned Dual Contouring (UDC) as the mesh representation. UDC discretizes a mesh in a regular grid and divides it into the face and vertex parts, recovering both complex structures and fine details. As a result, the one-to-one mapping between UDC and mesh resolves the ambiguity problem. In addition, GenUDC adopts a two-stage, coarse-to-fine generative process for 3D mesh generation. It first generates the face part as a rough shape and then the vertex part to craft a detailed shape. Extensive evaluations demonstrate the superiority of UDC as a mesh representation and the favorable performance of GenUDC in mesh generation. The code and trained models are available at https://github.com/TrepangCat/GenUDC.

comment: ACMMM 2024, code:https://github.com/TrepangCat/GenUDC

☆ TranSPORTmer: A Holistic Approach to Trajectory Understanding in Multi-Agent Sports ACCV 2024

Guillem Capellera, Luis Ferraz, Antonio Rubio, Antonio Agudo, Francesc Moreno-Noguer

Understanding trajectories in multi-agent scenarios requires addressing various tasks, including predicting future movements, imputing missing observations, inferring the status of unseen agents, and classifying different global states. Traditional data-driven approaches often handle these tasks separately with specialized models. We introduce TranSPORTmer, a unified transformer-based framework capable of addressing all these tasks, showcasing its application to the intricate dynamics of multi-agent sports scenarios like soccer and basketball. Using Set Attention Blocks, TranSPORTmer effectively captures temporal dynamics and social interactions in an equivariant manner. The model's tasks are guided by an input mask that conceals missing or yet-to-be-predicted observations. Additionally, we introduce a CLS extra agent to classify states along soccer trajectories, including passes, possessions, uncontrolled states, and out-of-play intervals, contributing to an enhancement in modeling trajectories. Evaluations on soccer and basketball datasets show that TranSPORTmer outperforms state-of-the-art task-specific models in player forecasting, player forecasting-imputation, ball inference, and ball imputation. https://youtu.be/8VtSRm8oGoE

comment: Accepted to ACCV 2024

☆ ADEM-VL: Adaptive and Embedded Fusion for Efficient Vision-Language Tuning

Zhiwei Hao, Jianyuan Guo, Li Shen, Yong Luo, Han Hu, Yonggang Wen

Recent advancements in multimodal fusion have witnessed the remarkable success of vision-language (VL) models, which excel in various multimodal applications such as image captioning and visual question answering. However, building VL models requires substantial hardware resources, where efficiency is restricted by two key factors: the extended input sequence of the language model with vision features demands more computational operations, and a large number of additional learnable parameters increase memory complexity. These challenges significantly restrict the broader applicability of such models. To bridge this gap, we propose ADEM-VL, an efficient vision-language method that tunes VL models based on pretrained large language models (LLMs) by adopting a parameter-free cross-attention mechanism for similarity measurements in multimodal fusion. This approach only requires embedding vision features into the language space, significantly reducing the number of trainable parameters and accelerating both training and inference speeds. To enhance representation learning in fusion module, we introduce an efficient multiscale feature generation scheme that requires only a single forward pass through the vision encoder. Moreover, we propose an adaptive fusion scheme that dynamically discards less relevant visual information for each text token based on its attention score. This ensures that the fusion process prioritizes the most pertinent visual features. With experiments on various tasks including visual question answering, image captioning, and instruction-following, we demonstrate that our framework outperforms existing approaches. Specifically, our method surpasses existing methods by an average accuracy of 0.77% on ScienceQA dataset, with reduced training and inference latency, demonstrating the superiority of our framework. The code is available at https://github.com/Hao840/ADEM-VL.

☆ Quasi-Medial Distance Field (Q-MDF): A Robust Method for Approximating and Discretizing Neural Medial Axis

Jiayi Kong, Chen Zong, Jun Luo, Shiqing Xin, Fei Hou, Hanqing Jiang, Chen Qian, Ying He

The medial axis, a lower-dimensional shape descriptor, plays an important role in the field of digital geometry processing. Despite its importance, robust computation of the medial axis transform from diverse inputs, especially point clouds with defects, remains a significant challenge. In this paper, we tackle the challenge by proposing a new implicit method that diverges from mainstream explicit medial axis computation techniques. Our key technical insight is the difference between the signed distance field (SDF) and the medial field (MF) of a solid shape is the unsigned distance field (UDF) of the shape's medial axis. This allows for formulating medial axis computation as an implicit reconstruction problem. Utilizing a modified double covering method, we extract the medial axis as the zero level-set of the UDF. Extensive experiments show that our method has enhanced accuracy and robustness in learning compact medial axis transform from thorny meshes and point clouds compared to existing methods.

☆ Scaling Robot Policy Learning via Zero-Shot Labeling with Foundation Models

Nils Blank, Moritz Reuss, Marcel Rühle, Ömer Erdinç Yağmurlu, Fabian Wenzel, Oier Mees, Rudolf Lioutikov

A central challenge towards developing robots that can relate human language to their perception and actions is the scarcity of natural language annotations in diverse robot datasets. Moreover, robot policies that follow natural language instructions are typically trained on either templated language or expensive human-labeled instructions, hindering their scalability. To this end, we introduce NILS: Natural language Instruction Labeling for Scalability. NILS automatically labels uncurated, long-horizon robot data at scale in a zero-shot manner without any human intervention. NILS combines pretrained vision-language foundation models in order to detect objects in a scene, detect object-centric changes, segment tasks from large datasets of unlabelled interaction data and ultimately label behavior datasets. Evaluations on BridgeV2, Fractal, and a kitchen play dataset show that NILS can autonomously annotate diverse robot demonstrations of unlabeled and unstructured datasets while alleviating several shortcomings of crowdsourced human annotations, such as low data quality and diversity. We use NILS to label over 115k trajectories obtained from over 430 hours of robot data. We open-source our auto-labeling code and generated annotations on our website: http://robottasklabeling.github.io.

comment: Project Website at https://robottasklabeling.github.io/

☆ AdaDiffSR: Adaptive Region-aware Dynamic Acceleration Diffusion Model for Real-World Image Super-Resolution ECCV2024

Yuanting Fan, Chengxu Liu, Nengzhong Yin, Changlong Gao, Xueming Qian

Diffusion models (DMs) have shown promising results on single-image super-resolution and other image-to-image translation tasks. Benefiting from more computational resources and longer inference times, they are able to yield more realistic images. Existing DMs-based super-resolution methods try to achieve an overall average recovery over all regions via iterative refinement, ignoring the consideration that different input image regions require different timesteps to reconstruct. In this work, we notice that previous DMs-based super-resolution methods suffer from wasting computational resources to reconstruct invisible details. To further improve the utilization of computational resources, we propose AdaDiffSR, a DMs-based SR pipeline with dynamic timesteps sampling strategy (DTSS). Specifically, by introducing the multi-metrics latent entropy module (MMLE), we can achieve dynamic perception of the latent spatial information gain during the denoising process, thereby guiding the dynamic selection of the timesteps. In addition, we adopt a progressive feature injection module (PFJ), which dynamically injects the original image features into the denoising process based on the current information gain, so as to generate images with both fidelity and realism. Experiments show that our AdaDiffSR achieves comparable performance over current state-of-the-art DMs-based SR methods while consuming less computational resources and inference time on both synthetic and real-world datasets.

comment: 18 pages, 6 figures, ECCV2024 accepted

☆ VISAGE: Video Synthesis using Action Graphs for Surgery MICCAI 2024

Yousef Yeganeh, Rachmadio Lazuardi, Amir Shamseddin, Emine Dari, Yash Thirani, Nassir Navab Azade Farshad

Surgical data science (SDS) is a field that analyzes patient data before, during, and after surgery to improve surgical outcomes and skills. However, surgical data is scarce, heterogeneous, and complex, which limits the applicability of existing machine learning methods. In this work, we introduce the novel task of future video generation in laparoscopic surgery. This task can augment and enrich the existing surgical data and enable various applications, such as simulation, analysis, and robot-aided surgery. Ultimately, it involves not only understanding the current state of the operation but also accurately predicting the dynamic and often unpredictable nature of surgical procedures. Our proposed method, VISAGE (VIdeo Synthesis using Action Graphs for Surgery), leverages the power of action scene graphs to capture the sequential nature of laparoscopic procedures and utilizes diffusion models to synthesize temporally coherent video sequences. VISAGE predicts the future frames given only a single initial frame, and the action graph triplets. By incorporating domain-specific knowledge through the action graph, VISAGE ensures the generated videos adhere to the expected visual and motion patterns observed in real laparoscopic procedures. The results of our experiments demonstrate high-fidelity video generation for laparoscopy procedures, which enables various applications in SDS.

comment: Accepted at MICCAI 2024 Embodied AI and Robotics for HealTHcare (EARTH) Workshop

☆ Efficient Neural Implicit Representation for 3D Human Reconstruction

Zexu Huang, Sarah Monazam Erfani, Siying Lu, Mingming Gong

High-fidelity digital human representations are increasingly in demand in the digital world, particularly for interactive telepresence, AR/VR, 3D graphics, and the rapidly evolving metaverse. Even though they work well in small spaces, conventional methods for reconstructing 3D human motion frequently require the use of expensive hardware and have high processing costs. This study presents HumanAvatar, an innovative approach that efficiently reconstructs precise human avatars from monocular video sources. At the core of our methodology, we integrate the pre-trained HuMoR, a model celebrated for its proficiency in human motion estimation. This is adeptly fused with the cutting-edge neural radiance field technology, Instant-NGP, and the state-of-the-art articulated model, Fast-SNARF, to enhance the reconstruction fidelity and speed. By combining these two technologies, a system is created that can render quickly and effectively while also providing estimation of human pose parameters that are unmatched in accuracy. We have enhanced our system with an advanced posture-sensitive space reduction technique, which optimally balances rendering quality with computational efficiency. In our detailed experimental analysis using both artificial and real-world monocular videos, we establish the advanced performance of our approach. HumanAvatar consistently equals or surpasses contemporary leading-edge reconstruction techniques in quality. Furthermore, it achieves these complex reconstructions in minutes, a fraction of the time typically required by existing methods. Our models achieve a training speed that is 110X faster than that of State-of-The-Art (SoTA) NeRF-based models. Our technique performs noticeably better than SoTA dynamic human NeRF methods if given an identical runtime limit. HumanAvatar can provide effective visuals after only 30 seconds of training.

☆ Emotion Recognition with Facial Attention and Objective Activation Functions

Andrzej Miskow, Abdulrahman Altahhan

In this paper, we study the effect of introducing channel and spatial attention mechanisms, namely SEN-Net, ECA-Net, and CBAM, to existing CNN vision-based models such as VGGNet, ResNet, and ResNetV2 to perform the Facial Emotion Recognition task. We show that not only attention can significantly improve the performance of these models but also that combining them with a different activation function can further help increase the performance of these models.

☆ New Insight in Cervical Cancer Diagnosis Using Convolution Neural Network Architecture

Ach. Khozaimi, Wayan Firdaus Mahmudy

The Pap smear is a screening method for early cervical cancer diagnosis. The selection of the right optimizer in the convolutional neural network (CNN) model is key to the success of the CNN in image classification, including the classification of cervical cancer Pap smear images. In this study, stochastic gradient descent (SGD), RMSprop, Adam, AdaGrad, AdaDelta, Adamax, and Nadam optimizers were used to classify cervical cancer Pap smear images from the SipakMed dataset. Resnet-18, Resnet-34, and VGG-16 are the CNN architectures used in this study, and each architecture uses a transfer-learning model. Based on the test results, we conclude that the transfer learning model performs better on all CNNs and optimization techniques and that in the transfer learning model, the optimization has little influence on the training of the model. Adamax, with accuracy values of 72.8% and 66.8%, had the best accuracy for the VGG-16 and Resnet-18 architectures, respectively. Resnet-34 had 54.0%. This is 0.034% lower than Nadam. Overall, Adamax is a suitable optimizer for CNN in cervical cancer classification on Resnet-18, Resnet-34, and VGG-16 architectures. This study provides new insights into the configuration of CNN models for Pap smear image analysis.

☆ YOLO-Vehicle-Pro: A Cloud-Edge Collaborative Framework for Object Detection in Autonomous Driving under Adverse Weather Conditions

Xiguang Li, Jiafu Chen, Yunhe Sun, Na Lin, Ammar Hawbani, Liang Zhao

With the rapid advancement of autonomous driving technology, efficient and accurate object detection capabilities have become crucial factors in ensuring the safety and reliability of autonomous driving systems. However, in low-visibility environments such as hazy conditions, the performance of traditional object detection algorithms often degrades significantly, failing to meet the demands of autonomous driving. To address this challenge, this paper proposes two innovative deep learning models: YOLO-Vehicle and YOLO-Vehicle-Pro. YOLO-Vehicle is an object detection model tailored specifically for autonomous driving scenarios, employing multimodal fusion techniques to combine image and textual information for object detection. YOLO-Vehicle-Pro builds upon this foundation by introducing an improved image dehazing algorithm, enhancing detection performance in low-visibility environments. In addition to model innovation, this paper also designs and implements a cloud-edge collaborative object detection system, deploying models on edge devices and offloading partial computational tasks to the cloud in complex situations. Experimental results demonstrate that on the KITTI dataset, the YOLO-Vehicle-v1s model achieved 92.1% accuracy while maintaining a detection speed of 226 FPS and an inference time of 12ms, meeting the real-time requirements of autonomous driving. When processing hazy images, the YOLO-Vehicle-Pro model achieved a high accuracy of 82.3% mAP@50 on the Foggy Cityscapes dataset while maintaining a detection speed of 43 FPS.

☆ YOLOv11: An Overview of the Key Architectural Enhancements

Rahima Khanam, Muhammad Hussain

This study presents an architectural analysis of YOLOv11, the latest iteration in the YOLO (You Only Look Once) series of object detection models. We examine the models architectural innovations, including the introduction of the C3k2 (Cross Stage Partial with kernel size 2) block, SPPF (Spatial Pyramid Pooling - Fast), and C2PSA (Convolutional block with Parallel Spatial Attention) components, which contribute in improving the models performance in several ways such as enhanced feature extraction. The paper explores YOLOv11's expanded capabilities across various computer vision tasks, including object detection, instance segmentation, pose estimation, and oriented object detection (OBB). We review the model's performance improvements in terms of mean Average Precision (mAP) and computational efficiency compared to its predecessors, with a focus on the trade-off between parameter count and accuracy. Additionally, the study discusses YOLOv11's versatility across different model sizes, from nano to extra-large, catering to diverse application needs from edge devices to high-performance computing environments. Our research provides insights into YOLOv11's position within the broader landscape of object detection and its potential impact on real-time computer vision applications.

☆ Continual Learning on a Data Diet

Elif Ceren Gok Yildirim, Murat Onur Yildirim, Joaquin Vanschoren

Continual Learning (CL) methods usually learn from all available data. However, this is not the case in human cognition which efficiently focuses on key experiences while disregarding the redundant information. Similarly, not all data points in a dataset have equal potential; some can be more informative than others. This disparity may significantly impact the performance, as both the quality and quantity of samples directly influence the model's generalizability and efficiency. Drawing inspiration from this, we explore the potential of learning from important samples and present an empirical study for evaluating coreset selection techniques in the context of CL to stimulate research in this unexplored area. We train different continual learners on increasing amounts of selected samples and investigate the learning-forgetting dynamics by shedding light on the underlying mechanisms driving their improved stability-plasticity balance. We present several significant observations: learning from selectively chosen samples (i) enhances incremental accuracy, (ii) improves knowledge retention of previous tasks, and (iii) refines learned representations. This analysis contributes to a deeper understanding of selective learning strategies in CL scenarios.

comment: 18 pages, 6 figures

☆ Longitudinal Causal Image Synthesis

Yujia Li, Han Li, ans S. Kevin Zhou

Clinical decision-making relies heavily on causal reasoning and longitudinal analysis. For example, for a patient with Alzheimer's disease (AD), how will the brain grey matter atrophy in a year if intervened on the A-beta level in cerebrospinal fluid? The answer is fundamental to diagnosis and follow-up treatment. However, this kind of inquiry involves counterfactual medical images which can not be acquired by instrumental or correlation-based image synthesis models. Yet, such queries require counterfactual medical images, not obtainable through standard image synthesis models. Hence, a causal longitudinal image synthesis (CLIS) method, enabling the synthesis of such images, is highly valuable. However, building a CLIS model confronts three primary yet unmet challenges: mismatched dimensionality between high-dimensional images and low-dimensional tabular variables, inconsistent collection intervals of follow-up data, and inadequate causal modeling capability of existing causal graph methods for image data. In this paper, we established a tabular-visual causal graph (TVCG) for CLIS overcoming these challenges through a novel integration of generative imaging, continuous-time modeling, and structural causal models combined with a neural network. We train our CLIS based on the ADNI dataset and evaluate it on two other AD datasets, which illustrate the outstanding yet controllable quality of the synthesized images and the contributions of synthesized MRI to the characterization of AD progression, substantiating the reliability and utility in clinics.

☆ Deep Generative Models for 3D Medical Image Synthesis

Paul Friedrich, Yannik Frisch, Philippe C. Cattin

Deep generative modeling has emerged as a powerful tool for synthesizing realistic medical images, driving advances in medical image analysis, disease diagnosis, and treatment planning. This chapter explores various deep generative models for 3D medical image synthesis, with a focus on Variational Autoencoders (VAEs), Generative Adversarial Networks (GANs), and Denoising Diffusion Models (DDMs). We discuss the fundamental principles, recent advances, as well as strengths and weaknesses of these models and examine their applications in clinically relevant problems, including unconditional and conditional generation tasks like image-to-image translation and image reconstruction. We additionally review commonly used evaluation metrics for assessing image fidelity, diversity, utility, and privacy and provide an overview of current challenges in the field.

☆ Surgical Scene Segmentation by Transformer With Asymmetric Feature Enhancement

Cheng Yuan, Yutong Ban

Surgical scene segmentation is a fundamental task for robotic-assisted laparoscopic surgery understanding. It often contains various anatomical structures and surgical instruments, where similar local textures and fine-grained structures make the segmentation a difficult task. Vision-specific transformer method is a promising way for surgical scene understanding. However, there are still two main challenges. Firstly, the absence of inner-patch information fusion leads to poor segmentation performance. Secondly, the specific characteristics of anatomy and instruments are not specifically modeled. To tackle the above challenges, we propose a novel Transformer-based framework with an Asymmetric Feature Enhancement module (TAFE), which enhances local information and then actively fuses the improved feature pyramid into the embeddings from transformer encoders by a multi-scale interaction attention strategy. The proposed method outperforms the SOTA methods in several different surgical segmentation tasks and additionally proves its ability of fine-grained structure recognition. Code is available at https://github.com/cyuan-sjtu/ViT-asym.

☆ MIA-DPO: Multi-Image Augmented Direct Preference Optimization For Large Vision-Language Models

Ziyu Liu, Yuhang Zang, Xiaoyi Dong, Pan Zhang, Yuhang Cao, Haodong Duan, Conghui He, Yuanjun Xiong, Dahua Lin, Jiaqi Wang

Visual preference alignment involves training Large Vision-Language Models (LVLMs) to predict human preferences between visual inputs. This is typically achieved by using labeled datasets of chosen/rejected pairs and employing optimization algorithms like direct preference optimization (DPO). Existing visual alignment methods, primarily designed for single-image scenarios, struggle to effectively handle the complexity of multi-image tasks due to the scarcity of diverse training data and the high cost of annotating chosen/rejected pairs. We present Multi-Image Augmented Direct Preference Optimization (MIA-DPO), a visual preference alignment approach that effectively handles multi-image inputs. MIA-DPO mitigates the scarcity of diverse multi-image training data by extending single-image data with unrelated images arranged in grid collages or pic-in-pic formats, significantly reducing the costs associated with multi-image data annotations. Our observation reveals that attention values of LVLMs vary considerably across different images. We use attention values to identify and filter out rejected responses the model may have mistakenly focused on. Our attention-aware selection for constructing the chosen/rejected pairs without relying on (i) human annotation, (ii) extra data, and (iii) external models or APIs. MIA-DPO is compatible with various architectures and outperforms existing methods on five multi-image benchmarks, achieving an average performance boost of 3.0% on LLaVA-v1.5 and 4.3% on the recent InternLM-XC2.5. Moreover, MIA-DPO has a minimal effect on the model's ability to understand single images.

comment: Project URL: https://github.com/Liuziyu77/MIA-DPO

☆ Bridging the Gaps: Utilizing Unlabeled Face Recognition Datasets to Boost Semi-Supervised Facial Expression Recognition

Jie Song, Mengqiao He, Jinhua Feng, Bairong Shen

In recent years, Facial Expression Recognition (FER) has gained increasing attention. Most current work focuses on supervised learning, which requires a large amount of labeled and diverse images, while FER suffers from the scarcity of large, diverse datasets and annotation difficulty. To address these problems, we focus on utilizing large unlabeled Face Recognition (FR) datasets to boost semi-supervised FER. Specifically, we first perform face reconstruction pre-training on large-scale facial images without annotations to learn features of facial geometry and expression regions, followed by two-stage fine-tuning on FER datasets with limited labels. In addition, to further alleviate the scarcity of labeled and diverse images, we propose a Mixup-based data augmentation strategy tailored for facial images, and the loss weights of real and virtual images are determined according to the intersection-over-union (IoU) of the faces in the two images. Experiments on RAF-DB, AffectNet, and FERPlus show that our method outperforms existing semi-supervised FER methods and achieves new state-of-the-art performance. Remarkably, with only 5%, 25% training sets,our method achieves 64.02% on AffectNet,and 88.23% on RAF-DB, which is comparable to fully supervised state-of-the-art methods. Codes will be made publicly available at https://github.com/zhelishisongjie/SSFER.

☆ ImDy: Human Inverse Dynamics from Imitated Observations

Xinpeng Liu, Junxuan Liang, Zili Lin, Haowen Hou, Yong-Lu Li, Cewu Lu

Inverse dynamics (ID), which aims at reproducing the driven torques from human kinematic observations, has been a critical tool for gait analysis. However, it is hindered from wider application to general motion due to its limited scalability. Conventional optimization-based ID requires expensive laboratory setups, restricting its availability. To alleviate this problem, we propose to exploit the recently progressive human motion imitation algorithms to learn human inverse dynamics in a data-driven manner. The key insight is that the human ID knowledge is implicitly possessed by motion imitators, though not directly applicable. In light of this, we devise an efficient data collection pipeline with state-of-the-art motion imitation algorithms and physics simulators, resulting in a large-scale human inverse dynamics benchmark as Imitated Dynamics (ImDy). ImDy contains over 150 hours of motion with joint torque and full-body ground reaction force data. With ImDy, we train a data-driven human inverse dynamics solver ImDyS(olver) in a fully supervised manner, which conducts ID and ground reaction force estimation simultaneously. Experiments on ImDy and real-world data demonstrate the impressive competency of ImDyS in human inverse dynamics and ground reaction force estimation. Moreover, the potential of ImDy(-S) as a fundamental motion analysis tool is exhibited with downstream applications. The project page is https://foruck.github.io/ImDy/.

comment: Yong-Lu Li and Cewu Lu are the corresponding authors

☆ Towards Effective Data-Free Knowledge Distillation via Diverse Diffusion Augmentation

Muquan Li, Dongyang Zhang, Tao He, Xiurui Xie, Yuan-Fang Li, Ke Qin

Data-free knowledge distillation (DFKD) has emerged as a pivotal technique in the domain of model compression, substantially reducing the dependency on the original training data. Nonetheless, conventional DFKD methods that employ synthesized training data are prone to the limitations of inadequate diversity and discrepancies in distribution between the synthesized and original datasets. To address these challenges, this paper introduces an innovative approach to DFKD through diverse diffusion augmentation (DDA). Specifically, we revise the paradigm of common data synthesis in DFKD to a composite process through leveraging diffusion models subsequent to data synthesis for self-supervised augmentation, which generates a spectrum of data samples with similar distributions while retaining controlled variations. Furthermore, to mitigate excessive deviation in the embedding space, we introduce an image filtering technique grounded in cosine similarity to maintain fidelity during the knowledge distillation process. Comprehensive experiments conducted on CIFAR-10, CIFAR-100, and Tiny-ImageNet datasets showcase the superior performance of our method across various teacher-student network configurations, outperforming the contemporary state-of-the-art DFKD methods. Code will be available at:https://github.com/SLGSP/DDA.

☆ PlantCamo: Plant Camouflage Detection

Jinyu Yang, Qingwei Wang, Feng Zheng, Peng Chen, Aleš Leonardis, Deng-Ping Fan

Camouflaged Object Detection (COD) aims to detect objects with camouflaged properties. Although previous studies have focused on natural (animals and insects) and unnatural (artistic and synthetic) camouflage detection, plant camouflage has been neglected. However, plant camouflage plays a vital role in natural camouflage. Therefore, this paper introduces a new challenging problem of Plant Camouflage Detection (PCD). To address this problem, we introduce the PlantCamo dataset, which comprises 1,250 images with camouflaged plants representing 58 object categories in various natural scenes. To investigate the current status of plant camouflage detection, we conduct a large-scale benchmark study using 20+ cutting-edge COD models on the proposed dataset. Due to the unique characteristics of plant camouflage, including holes and irregular borders, we developed a new framework, named PCNet, dedicated to PCD. Our PCNet surpasses performance thanks to its multi-scale global feature enhancement and refinement. Finally, we discuss the potential applications and insights, hoping this work fills the gap in fine-grained COD research and facilitates further intelligent ecology research. All resources will be available on https://github.com/yjybuaa/PlantCamo.

☆ How to Continually Adapt Text-to-Image Diffusion Models for Flexible Customization? NeurIPS2024

Jiahua Dong, Wenqi Liang, Hongliu Li, Duzhen Zhang, Meng Cao, Henghui Ding, Salman Khan, Fahad Shahbaz Khan

Custom diffusion models (CDMs) have attracted widespread attention due to their astonishing generative ability for personalized concepts. However, most existing CDMs unreasonably assume that personalized concepts are fixed and cannot change over time. Moreover, they heavily suffer from catastrophic forgetting and concept neglect on old personalized concepts when continually learning a series of new concepts. To address these challenges, we propose a novel Concept-Incremental text-to-image Diffusion Model (CIDM), which can resolve catastrophic forgetting and concept neglect to learn new customization tasks in a concept-incremental manner. Specifically, to surmount the catastrophic forgetting of old concepts, we develop a concept consolidation loss and an elastic weight aggregation module. They can explore task-specific and task-shared knowledge during training, and aggregate all low-rank weights of old concepts based on their contributions during inference. Moreover, in order to address concept neglect, we devise a context-controllable synthesis strategy that leverages expressive region features and noise estimation to control the contexts of generated images according to user conditions. Experiments validate that our CIDM surpasses existing custom diffusion models. The source codes are available at https://github.com/JiahuaDong/CIFC.

comment: Accepted to NeurIPS2024

☆ Double Banking on Knowledge: Customized Modulation and Prototypes for Multi-Modality Semi-supervised Medical Image Segmentation

Yingyu Chen, Ziyuan Yang, Ming Yan, Zhongzhou Zhang, Hui Yu, Yan Liu, Yi Zhang

Multi-modality (MM) semi-supervised learning (SSL) based medical image segmentation has recently gained increasing attention for its ability to utilize MM data and reduce reliance on labeled images. However, current methods face several challenges: (1) Complex network designs hinder scalability to scenarios with more than two modalities. (2) Focusing solely on modality-invariant representation while neglecting modality-specific features, leads to incomplete MM learning. (3) Leveraging unlabeled data with generative methods can be unreliable for SSL. To address these problems, we propose Double Bank Dual Consistency (DBDC), a novel MM-SSL approach for medical image segmentation. To address challenge (1), we propose a modality all-in-one segmentation network that accommodates data from any number of modalities, removing the limitation on modality count. To address challenge (2), we design two learnable plug-in banks, Modality-Level Modulation bank (MLMB) and Modality-Level Prototype (MLPB) bank, to capture both modality-invariant and modality-specific knowledge. These banks are updated using our proposed Modality Prototype Contrastive Learning (MPCL). Additionally, we design Modality Adaptive Weighting (MAW) to dynamically adjust learning weights for each modality, ensuring balanced MM learning as different modalities learn at different rates. Finally, to address challenge (3), we introduce a Dual Consistency (DC) strategy that enforces consistency at both the image and feature levels without relying on generative methods. We evaluate our method on a 2-to-4 modality segmentation task using three open-source datasets, and extensive experiments show that our method outperforms state-of-the-art approaches.

☆ BlurryScope: a cost-effective and compact scanning microscope for automated HER2 scoring using deep learning on blurry image data

Michael John Fanous, Christopher Michael Seybold, Hanlong Chen, Nir Pillar, Aydogan Ozcan

We developed a rapid scanning optical microscope, termed "BlurryScope", that leverages continuous image acquisition and deep learning to provide a cost-effective and compact solution for automated inspection and analysis of tissue sections. BlurryScope integrates specialized hardware with a neural network-based model to quickly process motion-blurred histological images and perform automated pathology classification. This device offers comparable speed to commercial digital pathology scanners, but at a significantly lower price point and smaller size/weight, making it ideal for fast triaging in small clinics, as well as for resource-limited settings. To demonstrate the proof-of-concept of BlurryScope, we implemented automated classification of human epidermal growth factor receptor 2 (HER2) scores on immunohistochemically (IHC) stained breast tissue sections, achieving concordant results with those obtained from a high-end digital scanning microscope. We evaluated this approach by scanning HER2-stained tissue microarrays (TMAs) at a continuous speed of 5 mm/s, which introduces bidirectional motion blur artifacts. These compromised images were then used to train our network models. Using a test set of 284 unique patient cores, we achieved blind testing accuracies of 79.3% and 89.7% for 4-class (0, 1+, 2+, 3+) and 2-class (0/1+ , 2+/3+) HER2 score classification, respectively. BlurryScope automates the entire workflow, from image scanning to stitching and cropping of regions of interest, as well as HER2 score classification. We believe BlurryScope has the potential to enhance the current pathology infrastructure in resource-scarce environments, save diagnostician time and bolster cancer identification and classification across various clinical environments.

comment: 18 Pages, 6 Figures

☆ Unsupervised Low-dose CT Reconstruction with One-way Conditional Normalizing Flows

Ran An, Ke Chen, Hongwei Li

Deep-learning methods have shown promising performance for low-dose computed tomography (LDCT) reconstruction. However, supervised methods face the problem of lacking labeled data in clinical scenarios, and the CNN-based unsupervised denoising methods would cause excessive smoothing in the reconstructed image. Recently, the normalizing flows (NFs) based methods have shown advantages in producing detail-rich images and avoiding over-smoothing, however, there are still issues: (1) Although the alternating optimization in the data and latent space can well utilize the regularization and generation capabilities of NFs, the current two-way transformation strategy of noisy images and latent variables would cause detail loss and secondary artifacts; and (2) Training NFs on high-resolution CT images is hard due to huge computation. Though using conditional normalizing flows (CNFs) to learn conditional probability can reduce the computational burden, current methods require labeled data for conditionalization, and the unsupervised CNFs-based LDCT reconstruction remains a problem. To tackle these problems, we propose a novel CNFs-based unsupervised LDCT iterative reconstruction algorithm. It employs strict one-way transformation when performing alternating optimization in the dual spaces, thus effectively avoiding the problems of detail loss and secondary artifacts. By proposing a novel unsupervised conditionalization strategy, we train CNFs on high-resolution CT images, thus achieving fast and high-quality unsupervised reconstruction. Experiments on different datasets suggest that the performance of the proposed algorithm could surpass some state-of-the-art unsupervised and even supervised methods.

☆ OVT-B: A New Large-Scale Benchmark for Open-Vocabulary Multi-Object Tracking NeurIPS 2024

Haiji Liang, Ruize Han

Open-vocabulary object perception has become an important topic in artificial intelligence, which aims to identify objects with novel classes that have not been seen during training. Under this setting, open-vocabulary object detection (OVD) in a single image has been studied in many literature. However, open-vocabulary object tracking (OVT) from a video has been studied less, and one reason is the shortage of benchmarks. In this work, we have built a new large-scale benchmark for open-vocabulary multi-object tracking namely OVT-B. OVT-B contains 1,048 categories of objects and 1,973 videos with 637,608 bounding box annotations, which is much larger than the sole open-vocabulary tracking dataset, i.e., OVTAO-val dataset (200+ categories, 900+ videos). The proposed OVT-B can be used as a new benchmark to pave the way for OVT research. We also develop a simple yet effective baseline method for OVT. It integrates the motion features for object tracking, which is an important feature for MOT but is ignored in previous OVT methods. Experimental results have verified the usefulness of the proposed benchmark and the effectiveness of our method. We have released the benchmark to the public at https://github.com/Coo1Sea/OVT-B-Dataset.

comment: 15 pages, 6 figures, accepted at NeurIPS 2024 Dataset and Benchmark Track

☆ Diffusion Priors for Variational Likelihood Estimation and Image Denoising NeurIPS2024

Jun Cheng, Shan Tan

Real-world noise removal is crucial in low-level computer vision. Due to the remarkable generation capabilities of diffusion models, recent attention has shifted towards leveraging diffusion priors for image restoration tasks. However, existing diffusion priors-based methods either consider simple noise types or rely on approximate posterior estimation, limiting their effectiveness in addressing structured and signal-dependent noise commonly found in real-world images. In this paper, we build upon diffusion priors and propose adaptive likelihood estimation and MAP inference during the reverse diffusion process to tackle real-world noise. We introduce an independent, non-identically distributed likelihood combined with the noise precision (inverse variance) prior and dynamically infer the precision posterior using variational Bayes during the generation process. Meanwhile, we rectify the estimated noise variance through local Gaussian convolution. The final denoised image is obtained by propagating intermediate MAP solutions that balance the updated likelihood and diffusion prior. Additionally, we explore the local diffusion prior inherent in low-resolution diffusion models, enabling direct handling of high-resolution noisy images. Extensive experiments and analyses on diverse real-world datasets demonstrate the effectiveness of our method. Code is available at https://github.com/HUST-Tan/DiffusionVI.

comment: Accepted by NeurIPS2024 as Spotlight

☆ PathMoCo: A Novel Framework to Improve Feature Embedding in Self-supervised Contrastive Learning for Histopathological Images

Hamid Manoochehri, Bodong Zhang, Beatrice S. Knudsen, Tolga Tasdizen

Self-supervised contrastive learning has become a cornerstone in various areas, particularly histopathological image analysis. Image augmentation plays a crucial role in self-supervised contrastive learning, as it generates variations in image samples. However, traditional image augmentation techniques often overlook the unique characteristics of histopathological images. In this paper, we propose a new histopathology-specific image augmentation method called stain reconstruction augmentation (SRA). We integrate our SRA with MoCo v3, a leading model in self-supervised contrastive learning, along with our additional contrastive loss terms, and call the new model PathMoCo. We demonstrate that our PathMoCo always outperforms the standard MoCo v3 across various downstream tasks and achieves comparable or superior performance to other foundation models pre-trained on significantly larger histopathology datasets.

☆ HCDN: A Change Detection Network for Construction Housekeeping Using Feature Fusion and Large Vision Models

Kailai Sun, Zherui Shao, Yang Miang Goh, Jing Tian, Vincent J. L. Gan

Workplace safety has received increasing attention as millions of workers worldwide suffer from work-related accidents. Despite poor housekeeping is a significant contributor to construction accidents, there remains a significant lack of technological research focused on improving housekeeping practices in construction sites. Recognizing and locating poor housekeeping in a dynamic construction site is an important task that can be improved through computer vision approaches. Despite advances in AI and computer vision, existing methods for detecting poor housekeeping conditions face many challenges, including limited explanations, lack of locating of poor housekeeping, and lack of annotated datasets. On the other hand, change detection which aims to detect the changed environmental conditions (e.g., changing from good to poor housekeeping) and 'where' the change has occurred (e.g., location of objects causing poor housekeeping), has not been explored to the problem of housekeeping management. To address these challenges, we propose the Housekeeping Change Detection Network (HCDN), an advanced change detection neural network that integrates a feature fusion module and a large vision model, achieving state-of-the-art performance. Additionally, we introduce the approach to establish a novel change detection dataset (named Housekeeping-CCD) focused on housekeeping in construction sites, along with a housekeeping segmentation dataset. Our contributions include significant performance improvements compared to existing methods, providing an effective tool for enhancing construction housekeeping and safety. To promote further development, we share our source code and trained models for global researchers: https://github.com/NUS-DBE/Housekeeping-CD.

☆ PLGS: Robust Panoptic Lifting with 3D Gaussian Splatting

Yu Wang, Xiaobao Wei, Ming Lu, Guoliang Kang

Previous methods utilize the Neural Radiance Field (NeRF) for panoptic lifting, while their training and rendering speed are unsatisfactory. In contrast, 3D Gaussian Splatting (3DGS) has emerged as a prominent technique due to its rapid training and rendering speed. However, unlike NeRF, the conventional 3DGS may not satisfy the basic smoothness assumption as it does not rely on any parameterized structures to render (e.g., MLPs). Consequently, the conventional 3DGS is, in nature, more susceptible to noisy 2D mask supervision. In this paper, we propose a new method called PLGS that enables 3DGS to generate consistent panoptic segmentation masks from noisy 2D segmentation masks while maintaining superior efficiency compared to NeRF-based methods. Specifically, we build a panoptic-aware structured 3D Gaussian model to introduce smoothness and design effective noise reduction strategies. For the semantic field, instead of initialization with structure from motion, we construct reliable semantic anchor points to initialize the 3D Gaussians. We then use these anchor points as smooth regularization during training. Additionally, we present a self-training approach using pseudo labels generated by merging the rendered masks with the noisy masks to enhance the robustness of PLGS. For the instance field, we project the 2D instance masks into 3D space and match them with oriented bounding boxes to generate cross-view consistent instance masks for supervision. Experiments on various benchmarks demonstrate that our method outperforms previous state-of-the-art methods in terms of both segmentation quality and speed.

☆ Bilateral Hippocampi Segmentation in Low Field MRIs Using Mutual Feature Learning via Dual-Views

Himashi Peiris, Zhaolin Chen

Accurate hippocampus segmentation in brain MRI is critical for studying cognitive and memory functions and diagnosing neurodevelopmental disorders. While high-field MRIs provide detailed imaging, low-field MRIs are more accessible and cost-effective, which eliminates the need for sedation in children, though they often suffer from lower image quality. In this paper, we present a novel deep-learning approach for the automatic segmentation of bilateral hippocampi in low-field MRIs. Extending recent advancements in infant brain segmentation to underserved communities through the use of low-field MRIs ensures broader access to essential diagnostic tools, thereby supporting better healthcare outcomes for all children. Inspired by our previous work, Co-BioNet, the proposed model employs a dual-view structure to enable mutual feature learning via high-frequency masking, enhancing segmentation accuracy by leveraging complementary information from different perspectives. Extensive experiments demonstrate that our method provides reliable segmentation outcomes for hippocampal analysis in low-resource settings. The code is publicly available at: https://github.com/himashi92/LoFiHippSeg.

☆ Enhancing Multimodal Medical Image Classification using Cross-Graph Modal Contrastive Learning

Jun-En Ding, Chien-Chin Hsu, Feng Liu

The classification of medical images is a pivotal aspect of disease diagnosis, often enhanced by deep learning techniques. However, traditional approaches typically focus on unimodal medical image data, neglecting the integration of diverse non-image patient data. This paper proposes a novel Cross-Graph Modal Contrastive Learning (CGMCL) framework for multimodal medical image classification. The model effectively integrates both image and non-image data by constructing cross-modality graphs and leveraging contrastive learning to align multimodal features in a shared latent space. An inter-modality feature scaling module further optimizes the representation learning process by reducing the gap between heterogeneous modalities. The proposed approach is evaluated on two datasets: a Parkinson's disease (PD) dataset and a public melanoma dataset. Results demonstrate that CGMCL outperforms conventional unimodal methods in accuracy, interpretability, and early disease prediction. Additionally, the method shows superior performance in multi-class melanoma classification. The CGMCL framework provides valuable insights into medical image classification while offering improved disease interpretability and predictive capabilities.

☆ Unsupervised Domain Adaptation for Action Recognition via Self-Ensembling and Conditional Embedding Alignment

Indrajeet Ghosh, Garvit Chugh, Abu Zaher Md Faridee, Nirmalya Roy

Recent advancements in deep learning-based wearable human action recognition (wHAR) have improved the capture and classification of complex motions, but adoption remains limited due to the lack of expert annotations and domain discrepancies from user variations. Limited annotations hinder the model's ability to generalize to out-of-distribution samples. While data augmentation can improve generalizability, unsupervised augmentation techniques must be applied carefully to avoid introducing noise. Unsupervised domain adaptation (UDA) addresses domain discrepancies by aligning conditional distributions with labeled target samples, but vanilla pseudo-labeling can lead to error propagation. To address these challenges, we propose $\mu$DAR, a novel joint optimization architecture comprised of three functions: (i) consistency regularizer between augmented samples to improve model classification generalizability, (ii) temporal ensemble for robust pseudo-label generation and (iii) conditional distribution alignment to improve domain generalizability. The temporal ensemble works by aggregating predictions from past epochs to smooth out noisy pseudo-label predictions, which are then used in the conditional distribution alignment module to minimize kernel-based class-wise conditional maximum mean discrepancy ($k$CMMD) between the source and target feature space to learn a domain invariant embedding. The consistency-regularized augmentations ensure that multiple augmentations of the same sample share the same labels; this results in (a) strong generalization with limited source domain samples and (b) consistent pseudo-label generation in target samples. The novel integration of these three modules in $\mu$DAR results in a range of $\approx$ 4-12% average macro-F1 score improvement over six state-of-the-art UDA methods in four benchmark wHAR datasets

comment: This work has been accepted to the Proceedings of the IEEE International Conference on Data Mining, 2024

☆ GenDP: 3D Semantic Fields for Category-Level Generalizable Diffusion Policy

Yixuan Wang, Guang Yin, Binghao Huang, Tarik Kelestemur, Jiuguang Wang, Yunzhu Li

Diffusion-based policies have shown remarkable capability in executing complex robotic manipulation tasks but lack explicit characterization of geometry and semantics, which often limits their ability to generalize to unseen objects and layouts. To enhance the generalization capabilities of Diffusion Policy, we introduce a novel framework that incorporates explicit spatial and semantic information via 3D semantic fields. We generate 3D descriptor fields from multi-view RGBD observations with large foundational vision models, then compare these descriptor fields against reference descriptors to obtain semantic fields. The proposed method explicitly considers geometry and semantics, enabling strong generalization capabilities in tasks requiring category-level generalization, resolving geometric ambiguities, and attention to subtle geometric details. We evaluate our method across eight tasks involving articulated objects and instances with varying shapes and textures from multiple object categories. Our method demonstrates its effectiveness by increasing Diffusion Policy's average success rate on unseen instances from 20% to 93%. Additionally, we provide a detailed analysis and visualization to interpret the sources of performance gain and explain how our method can generalize to novel instances.

comment: Accepted to Conference on Robot Learning (CoRL 2024). Project Page: https://robopil.github.io/GenDP/

☆ Which Client is Reliable?: A Reliable and Personalized Prompt-based Federated Learning for Medical Image Question Answering

He Zhu, Ren Togo, Takahiro Ogawa, Miki Haseyama

Conventional medical artificial intelligence (AI) models face barriers in clinical application and ethical issues owing to their inability to handle the privacy-sensitive characteristics of medical data. We present a novel personalized federated learning (pFL) method for medical visual question answering (VQA) models, addressing privacy reliability challenges in the medical domain. Our method introduces learnable prompts into a Transformer architecture to efficiently train it on diverse medical datasets without massive computational costs. Then we introduce a reliable client VQA model that incorporates Dempster-Shafer evidence theory to quantify uncertainty in predictions, enhancing the model's reliability. Furthermore, we propose a novel inter-client communication mechanism that uses maximum likelihood estimation to balance accuracy and uncertainty, fostering efficient integration of insights across clients.

♻ ☆ Toward Fairer Face Recognition Datasets

Alexandre Fournier-Montgieux, Michael Soumm, Adrian Popescu, Bertrand Luvison, Hervé Le Borgne

Face recognition and verification are two computer vision tasks whose performance has progressed with the introduction of deep representations. However, ethical, legal, and technical challenges due to the sensitive character of face data and biases in real training datasets hinder their development. Generative AI addresses privacy by creating fictitious identities, but fairness problems persist. We promote fairness by introducing a demographic attributes balancing mechanism in generated training datasets. We experiment with an existing real dataset, three generated training datasets, and the balanced versions of a diffusion-based dataset. We propose a comprehensive evaluation that considers accuracy and fairness equally and includes a rigorous regression-based statistical analysis of attributes. The analysis shows that balancing reduces demographic unfairness. Also, a performance gap persists despite generation becoming more accurate with time. The proposed balancing method and comprehensive verification evaluation promote fairer and transparent face recognition and verification.

♻ ☆ Meteor: Mamba-based Traversal of Rationale for Large Language and Vision Models

Byung-Kwan Lee, Chae Won Kim, Beomchan Park, Yong Man Ro

The rapid development of large language and vision models (LLVMs) has been driven by advances in visual instruction tuning. Recently, open-source LLVMs have curated high-quality visual instruction tuning datasets and utilized additional vision encoders or multiple computer vision models in order to narrow the performance gap with powerful closed-source LLVMs. These advancements are attributed to multifaceted information required for diverse capabilities, including fundamental image understanding, real-world knowledge about common-sense and non-object concepts (e.g., charts, diagrams, symbols, signs, and math problems), and step-by-step procedures for solving complex questions. Drawing from the multifaceted information, we present a new efficient LLVM, Mamba-based traversal of rationales (Meteor), which leverages multifaceted rationale to enhance understanding and answering capabilities. To embed lengthy rationales containing abundant information, we employ the Mamba architecture, capable of processing sequential data with linear time complexity. We introduce a new concept of traversal of rationale that facilitates efficient embedding of rationale. Subsequently, the backbone multimodal language model (MLM) is trained to generate answers with the aid of rationale. Through these steps, Meteor achieves significant improvements in vision language performances across multiple evaluation benchmarks requiring diverse capabilities, without scaling up the model size or employing additional vision encoders and computer vision models.

comment: Code is available in https://github.com/ByungKwanLee/Meteor

♻ ☆ Pruning By Explaining Revisited: Optimizing Attribution Methods to Prune CNNs and Transformers ECCV 2024

Sayed Mohammad Vakilzadeh Hatefi, Maximilian Dreyer, Reduan Achtibat, Thomas Wiegand, Wojciech Samek, Sebastian Lapuschkin

To solve ever more complex problems, Deep Neural Networks are scaled to billions of parameters, leading to huge computational costs. An effective approach to reduce computational requirements and increase efficiency is to prune unnecessary components of these often over-parameterized networks. Previous work has shown that attribution methods from the field of eXplainable AI serve as effective means to extract and prune the least relevant network components in a few-shot fashion. We extend the current state by proposing to explicitly optimize hyperparameters of attribution methods for the task of pruning, and further include transformer-based networks in our analysis. Our approach yields higher model compression rates of large transformer- and convolutional architectures (VGG, ResNet, ViT) compared to previous works, while still attaining high performance on ImageNet classification tasks. Here, our experiments indicate that transformers have a higher degree of over-parameterization compared to convolutional neural networks. Code is available at https://github.com/erfanhatefi/Pruning-by-eXplaining-in-PyTorch.

comment: Accepted as a workshop paper at ECCV 2024, 26 pages (11 pages manuscript, 3 pages references, 12 pages appendix)

♻ ☆ VILA-U: a Unified Foundation Model Integrating Visual Understanding and Generation

Yecheng Wu, Zhuoyang Zhang, Junyu Chen, Haotian Tang, Dacheng Li, Yunhao Fang, Ligeng Zhu, Enze Xie, Hongxu Yin, Li Yi, Song Han, Yao Lu

VILA-U is a Unified foundation model that integrates Video, Image, Language understanding and generation. Traditional visual language models (VLMs) use separate modules for understanding and generating visual content, which can lead to misalignment and increased complexity. In contrast, VILA-U employs a single autoregressive next-token prediction framework for both tasks, eliminating the need for additional components like diffusion models. This approach not only simplifies the model but also achieves near state-of-the-art performance in visual language understanding and generation. The success of VILA-U is attributed to two main factors: the unified vision tower that aligns discrete visual tokens with textual inputs during pretraining, which enhances visual perception, and autoregressive image generation can achieve similar quality as diffusion models with high-quality dataset. This allows VILA-U to perform comparably to more complex models using a fully token-based autoregressive framework.

comment: Code: https://github.com/mit-han-lab/vila-u. The first two authors contributed equally to this work

♻ ☆ JointMotion: Joint Self-Supervision for Joint Motion Prediction

Royden Wagner, Omer Sahin Tas, Marvin Klemp, Carlos Fernandez

We present JointMotion, a self-supervised pre-training method for joint motion prediction in self-driving vehicles. Our method jointly optimizes a scene-level objective connecting motion and environments, and an instance-level objective to refine learned representations. Scene-level representations are learned via non-contrastive similarity learning of past motion sequences and environment context. At the instance level, we use masked autoencoding to refine multimodal polyline representations. We complement this with an adaptive pre-training decoder that enables JointMotion to generalize across different environment representations, fusion mechanisms, and dataset characteristics. Notably, our method reduces the joint final displacement error of Wayformer, HPTR, and Scene Transformer models by 3\%, 8\%, and 12\%, respectively; and enables transfer learning between the Waymo Open Motion and the Argoverse 2 Motion Forecasting datasets. Code: https://github.com/kit-mrt/future-motion

comment: CoRL'24 camera-ready

♻ ☆ Telling Stories for Common Sense Zero-Shot Action Recognition ACCV 2024

Shreyank N Gowda, Laura Sevilla-Lara

Video understanding has long suffered from reliance on large labeled datasets, motivating research into zero-shot learning. Recent progress in language modeling presents opportunities to advance zero-shot video analysis, but constructing an effective semantic space relating action classes remains challenging. We address this by introducing a novel dataset, Stories, which contains rich textual descriptions for diverse action classes extracted from WikiHow articles. For each class, we extract multi-sentence narratives detailing the necessary steps, scenes, objects, and verbs that characterize the action. This contextual data enables modeling of nuanced relationships between actions, paving the way for zero-shot transfer. We also propose an approach that harnesses Stories to improve feature generation for training zero-shot classification. Without any target dataset fine-tuning, our method achieves new state-of-the-art on multiple benchmarks, improving top-1 accuracy by up to 6.1%. We believe Stories provides a valuable resource that can catalyze progress in zero-shot action recognition. The textual narratives forge connections between seen and unseen classes, overcoming the bottleneck of labeled data that has long impeded advancements in this exciting domain. The data can be found here: https://github.com/kini5gowda/Stories .

comment: Accepted in ACCV 2024!

♻ ☆ Generalizable Prompt Tuning for Vision-Language Models

Qian Zhang

Prompt tuning for vision-language models such as CLIP involves optimizing the text prompts used to generate image-text pairs for specific downstream tasks. While hand-crafted or template-based prompts are generally applicable to a wider range of unseen classes, they tend to perform poorly in downstream tasks (i.e., seen classes). Learnable soft prompts, on the other hand, often perform well in downstream tasks but lack generalizability. Additionally, prior research has predominantly concentrated on the textual modality, with very few studies attempting to explore the prompt's generalization potential from the visual modality. Keeping these limitations in mind, we investigate how to prompt tuning to obtain both a competitive downstream performance and generalization. The study shows that by treating soft and hand-crafted prompts as dual views of the textual modality, and maximizing their mutual information, we can better ensemble task-specific and general semantic information. Moreover, to generate more expressive prompts, the study introduces a class-wise augmentation from the visual modality, resulting in significant robustness to a wider range of unseen classes. Extensive evaluations on several benchmarks report that the proposed approach achieves competitive results in terms of both task-specific performance and general abilities.

comment: in progress

♻ ☆ Exploring the Adversarial Robustness of CLIP for AI-generated Image Detection

Vincenzo De Rosa, Fabrizio Guillaro, Giovanni Poggi, Davide Cozzolino, Luisa Verdoliva

In recent years, many forensic detectors have been proposed to detect AI-generated images and prevent their use for malicious purposes. Convolutional neural networks (CNNs) have long been the dominant architecture in this field and have been the subject of intense study. However, recently proposed Transformer-based detectors have been shown to match or even outperform CNN-based detectors, especially in terms of generalization. In this paper, we study the adversarial robustness of AI-generated image detectors, focusing on Contrastive Language-Image Pretraining (CLIP)-based methods that rely on Visual Transformer (ViT) backbones and comparing their performance with CNN-based methods. We study the robustness to different adversarial attacks under a variety of conditions and analyze both numerical results and frequency-domain patterns. CLIP-based detectors are found to be vulnerable to white-box attacks just like CNN-based detectors. However, attacks do not easily transfer between CNN-based and CLIP-based methods. This is also confirmed by the different distribution of the adversarial noise patterns in the frequency domain. Overall, this analysis provides new insights into the properties of forensic detectors that can help to develop more effective strategies.

♻ ☆ Leveraging Hallucinations to Reduce Manual Prompt Dependency in Promptable Segmentation NeurIPS 2024

Jian Hu, Jiayi Lin, Junchi Yan, Shaogang Gong

Promptable segmentation typically requires instance-specific manual prompts to guide the segmentation of each desired object. To minimize such a need, task-generic promptable segmentation has been introduced, which employs a single task-generic prompt to segment various images of different objects in the same task. Current methods use Multimodal Large Language Models (MLLMs) to reason detailed instance-specific prompts from a task-generic prompt for improving segmentation accuracy. The effectiveness of this segmentation heavily depends on the precision of these derived prompts. However, MLLMs often suffer hallucinations during reasoning, resulting in inaccurate prompting. While existing methods focus on eliminating hallucinations to improve a model, we argue that MLLM hallucinations can reveal valuable contextual insights when leveraged correctly, as they represent pre-trained large-scale knowledge beyond individual images. In this paper, we utilize hallucinations to mine task-related information from images and verify its accuracy for enhancing precision of the generated prompts. Specifically, we introduce an iterative Prompt-Mask Cycle generation framework (ProMaC) with a prompt generator and a mask generator.The prompt generator uses a multi-scale chain of thought prompting, initially exploring hallucinations for extracting extended contextual knowledge on a test image.These hallucinations are then reduced to formulate precise instance-specific prompts, directing the mask generator to produce masks that are consistent with task semantics by mask semantic alignment. The generated masks iteratively induce the prompt generator to focus more on task-relevant image areas and reduce irrelevant hallucinations, resulting jointly in better prompts and masks. Experiments on 5 benchmarks demonstrate the effectiveness of ProMaC. Code given in https://lwpyh.github.io/ProMaC/.

comment: NeurIPS 2024

♻ ☆ LocoMotion: Learning Motion-Focused Video-Language Representations ACCV 2024

Hazel Doughty, Fida Mohammad Thoker, Cees G. M. Snoek

This paper strives for motion-focused video-language representations. Existing methods to learn video-language representations use spatial-focused data, where identifying the objects and scene is often enough to distinguish the relevant caption. We instead propose LocoMotion to learn from motion-focused captions that describe the movement and temporal progression of local object motions. We achieve this by adding synthetic motions to videos and using the parameters of these motions to generate corresponding captions. Furthermore, we propose verb-variation paraphrasing to increase the caption variety and learn the link between primitive motions and high-level verbs. With this, we are able to learn a motion-focused video-language representation. Experiments demonstrate our approach is effective for a variety of downstream tasks, particularly when limited data is available for fine-tuning. Code is available: https://hazeldoughty.github.io/Papers/LocoMotion/

comment: ACCV 2024 Oral

♻ ☆ Accessible, At-Home Detection of Parkinson's Disease via Multi-task Video Analysis

Md Saiful Islam, Tariq Adnan, Jan Freyberg, Sangwu Lee, Abdelrahman Abdelkader, Meghan Pawlik, Cathe Schwartz, Karen Jaffe, Ruth B. Schneider, E Ray Dorsey, Ehsan Hoque

Limited accessibility to neurological care leads to underdiagnosed Parkinson's Disease (PD), preventing early intervention. Existing AI-based PD detection methods primarily focus on unimodal analysis of motor or speech tasks, overlooking the multifaceted nature of the disease. To address this, we introduce a large-scale, multi-task video dataset consisting of 1102 sessions (each containing videos of finger tapping, facial expression, and speech tasks captured via webcam) from 845 participants (272 with PD). We propose a novel Uncertainty-calibrated Fusion Network (UFNet) that leverages this multimodal data to enhance diagnostic accuracy. UFNet employs independent task-specific networks, trained with Monte Carlo Dropout for uncertainty quantification, followed by self-attended fusion of features, with attention weights dynamically adjusted based on task-specific uncertainties. To ensure patient-centered evaluation, the participants were randomly split into three sets: 60% for training, 20% for model selection, and 20% for final performance evaluation. UFNet significantly outperformed single-task models in terms of accuracy, area under the ROC curve (AUROC), and sensitivity while maintaining non-inferior specificity. Withholding uncertain predictions further boosted the performance, achieving 88.0+-0.3%$ accuracy, 93.0+-0.2% AUROC, 79.3+-0.9% sensitivity, and 92.6+-0.3% specificity, at the expense of not being able to predict for 2.3+-0.3% data (+- denotes 95% confidence interval). Further analysis suggests that the trained model does not exhibit any detectable bias across sex and ethnic subgroups and is most effective for individuals aged between 50 and 80. Requiring only a webcam and microphone, our approach facilitates accessible home-based PD screening, especially in regions with limited healthcare resources.

♻ ☆ SCA: Highly Efficient Semantic-Consistent Unrestricted Adversarial Attack

Zihao Pan, Weibin Wu, Yuhang Cao, Zibin Zheng

Deep neural network based systems deployed in sensitive environments are vulnerable to adversarial attacks. Unrestricted adversarial attacks typically manipulate the semantic content of an image (e.g., color or texture) to create adversarial examples that are both effective and photorealistic. Recent works have utilized the diffusion inversion process to map images into a latent space, where high-level semantics are manipulated by introducing perturbations. However, they often results in substantial semantic distortions in the denoised output and suffers from low efficiency. In this study, we propose a novel framework called Semantic-Consistent Unrestricted Adversarial Attacks (SCA), which employs an inversion method to extract edit-friendly noise maps and utilizes Multimodal Large Language Model (MLLM) to provide semantic guidance throughout the process. Under the condition of rich semantic information provided by MLLM, we perform the DDPM denoising process of each step using a series of edit-friendly noise maps, and leverage DPM Solver++ to accelerate this process, enabling efficient sampling with semantic consistency. Compared to existing methods, our framework enables the efficient generation of adversarial examples that exhibit minimal discernible semantic changes. Consequently, we for the first time introduce Semantic-Consistent Adversarial Examples (SCAE). Extensive experiments and visualizations have demonstrated the high efficiency of SCA, particularly in being on average 12 times faster than the state-of-the-art attacks. Our research can further draw attention to the security of multimedia information.

♻ ☆ PnLCalib: Sports Field Registration via Points and Lines Optimization

Marc Gutiérrez-Pérez, Antonio Agudo

Camera calibration in broadcast sports videos presents numerous challenges for accurate sports field registration due to multiple camera angles, varying camera parameters, and frequent occlusions of the field. Traditional search-based methods depend on initial camera pose estimates, which can struggle in non-standard positions and dynamic environments. In response, we propose an optimization-based calibration pipeline that leverages a 3D soccer field model and a predefined set of keypoints to overcome these limitations. Our method also introduces a novel refinement module that improves initial calibration by using detected field lines in a non-linear optimization process. This approach outperforms existing techniques in both multi-view and single-view 3D camera calibration tasks, while maintaining competitive performance in homography estimation. Extensive experimentation on real-world soccer datasets, including SoccerNet-Calibration, WorldCup 2014, and TS-WorldCup, highlights the robustness and accuracy of our method across diverse broadcast scenarios. Our approach offers significant improvements in camera calibration precision and reliability.

comment: Extended version of "No Bells, Just Whistles: Sports Field Registration Leveraging Geometric Properties"

♻ ☆ PixLore: A Dataset-driven Approach to Rich Image Captioning

Diego Bonilla-Salvador, Marcelino Martínez-Sober, Joan Vila-Francés, Antonio José Serrano-López, Pablo Rodríguez-Belenguer, Fernando Mateo

In the domain of vision-language integration, generating detailed image captions poses a significant challenge due to the lack of curated and rich datasets. This study introduces PixLore, a novel method that leverages Querying Transformers through the fine-tuning of the BLIP-2 model using the LoRa method on a standard commercial GPU. The followed approach, which involves training on a carefully assembled dataset from state-of-the-art Computer Vision models combined and augmented by ChatGPT, addresses the question of whether intricate image understanding can be achieved with an ensemble of smaller-scale models, referred to as Knowledge Stitching. Comparative evaluations against major models such as GPT-4 and Google Bard demonstrate that PixLore-2.7B, despite having considerably fewer parameters, is rated higher than the existing State-of-the-Art models in over half of the assessments. Precisely, PixLore outperform Bard and BLIP-2, which score approximately 35.18% and 27.98% lower than PixLore in the task of image captioning. This research not only presents a groundbreaking approach but also highlights the importance of well-curated datasets in enhancing the performance of smaller models.

comment: Paper in preprint pending of publication

♻ ☆ Denoising Diffusion Models for Inpainting of Healthy Brain Tissue MICCAI

Alicia Durrer, Philippe C. Cattin, Julia Wolleb

This paper is a contribution to the "BraTS 2023 Local Synthesis of Healthy Brain Tissue via Inpainting Challenge". The task of this challenge is to transform tumor tissue into healthy tissue in brain magnetic resonance (MR) images. This idea originates from the problem that MR images can be evaluated using automatic processing tools, however, many of these tools are optimized for the analysis of healthy tissue. By solving the given inpainting task, we enable the automatic analysis of images featuring lesions, and further downstream tasks. Our approach builds on denoising diffusion probabilistic models. We use a 2D model that is trained using slices in which healthy tissue was cropped out and is learned to be inpainted again. This allows us to use the ground truth healthy tissue during training. In the sampling stage, we replace the slices containing diseased tissue in the original 3D volume with the slices containing the healthy tissue inpainting. With our approach, we achieve comparable results to the competing methods. On the validation set our model achieves a mean SSIM of 0.7804, a PSNR of 20.3525 and a MSE of 0.0113. In future we plan to extend our 2D model to a 3D model, allowing to inpaint the region of interest as a whole without losing context information of neighboring slices.

comment: 12 pages, 5 figures, MICCAI challenge submission

♻ ☆ A Multimodal Fusion Network For Student Emotion Recognition Based on Transformer and Tensor Product

Ao Xiang, Zongqing Qi, Han Wang, Qin Yang, Danqing Ma

This paper introduces a new multi-modal model based on the Transformer architecture and tensor product fusion strategy, combining BERT's text vectors and ViT's image vectors to classify students' psychological conditions, with an accuracy of 93.65%. The purpose of the study is to accurately analyze the mental health status of students from various data sources. This paper discusses modal fusion methods, including early, late and intermediate fusion, to overcome the challenges of integrating multi-modal information. Ablation studies compare the performance of different models and fusion techniques, showing that the proposed model outperforms existing methods such as CLIP and ViLBERT in terms of accuracy and inference speed. Conclusions indicate that while this model has significant advantages in emotion recognition, its potential to incorporate other data modalities provides areas for future research.

♻ ☆ Towards Croppable Implicit Neural Representations NeurIPS 2024

Maor Ashkenazi, Eran Treister

Implicit Neural Representations (INRs) have peaked interest in recent years due to their ability to encode natural signals using neural networks. While INRs allow for useful applications such as interpolating new coordinates and signal compression, their black-box nature makes it difficult to modify them post-training. In this paper we explore the idea of editable INRs, and specifically focus on the widely used cropping operation. To this end, we present Local-Global SIRENs -- a novel INR architecture that supports cropping by design. Local-Global SIRENs are based on combining local and global feature extraction for signal encoding. What makes their design unique is the ability to effortlessly remove specific portions of an encoded signal, with a proportional weight decrease. This is achieved by eliminating the corresponding weights from the network, without the need for retraining. We further show how this architecture can be used to support the straightforward extension of previously encoded signals. Beyond signal editing, we examine how the Local-Global approach can accelerate training, enhance encoding of various signals, improve downstream performance, and be applied to modern INRs such as INCODE, highlighting its potential and flexibility. Code is available at https://github.com/maorash/Local-Global-INRs.

comment: Accepted to NeurIPS 2024

♻ ☆ Breaking Class Barriers: Efficient Dataset Distillation via Inter-Class Feature Compensator

Xin Zhang, Jiawei Du, Ping Liu, Joey Tianyi Zhou

Dataset distillation has emerged as a technique aiming to condense informative features from large, natural datasets into a compact and synthetic form. While recent advancements have refined this technique, its performance is bottlenecked by the prevailing class-specific synthesis paradigm. Under this paradigm, synthetic data is optimized exclusively for a pre-assigned one-hot label, creating an implicit class barrier in feature condensation. This leads to inefficient utilization of the distillation budget and oversight of inter-class feature distributions, which ultimately limits the effectiveness and efficiency, as demonstrated in our analysis. To overcome these constraints, this paper presents the Inter-class Feature Compensator (INFER), an innovative distillation approach that transcends the class-specific data-label framework widely utilized in current dataset distillation methods. Specifically, INFER leverages a Universal Feature Compensator (UFC) to enhance feature integration across classes, enabling the generation of multiple additional synthetic instances from a single UFC input. This significantly improves the efficiency of the distillation budget. Moreover, INFER enriches inter-class interactions during the distillation, thereby enhancing the effectiveness and generalizability of the distilled data. By allowing for the linear interpolation of labels similar to those in the original dataset, INFER meticulously optimizes the synthetic data and dramatically reduces the size of soft labels in the synthetic dataset to almost zero, establishing a new benchmark for efficiency and effectiveness in dataset distillation.

♻ ☆ Exploring Stronger Transformer Representation Learning for Occluded Person Re-Identification

Zhangjian Ji, Donglin Cheng, Kai Feng

Due to some complex factors (e.g., occlusion, pose variation and diverse camera perspectives), extracting stronger feature representation in person re-identification remains a challenging task. In this paper, we proposed a novel self-supervision and supervision combining transformer-based person re-identification framework, namely SSSC-TransReID. Different from the general transformer-based person re-identification models, we designed a self-supervised contrastive learning branch, which can enhance the feature representation for person re-identification without negative samples or additional pre-training. In order to train the contrastive learning branch, we also proposed a novel random rectangle mask strategy to simulate the occlusion in real scenes, so as to enhance the feature representation for occlusion. Finally, we utilized the joint-training loss function to integrate the advantages of supervised learning with ID tags and self-supervised contrastive learning without negative samples, which can reinforce the ability of our model to excavate stronger discriminative features, especially for occlusion. Extensive experimental results on several benchmark datasets show our proposed model obtains superior Re-ID performance consistently and outperforms the state-of-the-art ReID methods by large margins on the mean average accuracy (mAP) and Rank-1 accuracy.

♻ ☆ From Real Artifacts to Virtual Reference: A Robust Framework for Translating Endoscopic Images

Junyang Wu, Fangfang Xie, Jiayuan Sun, Yun Gu, Guang-Zhong Yang

Domain adaptation, which bridges the distributions across different modalities, plays a crucial role in multimodal medical image analysis. In endoscopic imaging, combining pre-operative data with intra-operative imaging is important for surgical planning and navigation. However, existing domain adaptation methods are hampered by distribution shift caused by in vivo artifacts, necessitating robust techniques for aligning noisy and artifact abundant patient endoscopic videos with clean virtual images reconstructed from pre-operative tomographic data for pose estimation during intraoperative guidance. This paper presents an artifact-resilient image translation method and an associated benchmark for this purpose. The method incorporates a novel ``local-global'' translation framework and a noise-resilient feature extraction strategy. For the former, it decouples the image translation process into a local step for feature denoising, and a global step for global style transfer. For feature extraction, a new contrastive learning strategy is proposed, which can extract noise-resilient features for establishing robust correspondence across domains. Detailed validation on both public and in-house clinical datasets has been conducted, demonstrating significantly improved performance compared to the current state-of-the-art.

♻ ☆ Few-Shot Adversarial Prompt Learning on Vision-Language Models NeurIPS 2024

Yiwei Zhou, Xiaobo Xia, Zhiwei Lin, Bo Han, Tongliang Liu

The vulnerability of deep neural networks to imperceptible adversarial perturbations has attracted widespread attention. Inspired by the success of vision-language foundation models, previous efforts achieved zero-shot adversarial robustness by aligning adversarial visual features with text supervision. However, in practice, they are still unsatisfactory due to several issues, including heavy adaptation cost, suboptimal text supervision, and uncontrolled natural generalization capacity. In this paper, to address these issues, we propose a few-shot adversarial prompt framework where adapting input sequences with limited data makes significant adversarial robustness improvement. Specifically, we achieve this by providing adversarially correlated text supervision that is end-to-end learned from adversarial examples. We also propose a novel training objective that enhances the consistency of multi-modal features while encourages differentiated uni-modal features between natural and adversarial examples. The proposed framework gives access to learn adversarial text supervision, which provides superior cross-modal adversarial alignment and matches state-of-the-art zero-shot adversarial robustness with only 1% training data. Code is available at: https://github.com/lionel-w2/FAP.

comment: NeurIPS 2024

♻ ☆ Enhancing Interaction Modeling with Agent Selection and Physical Coefficient for Trajectory Prediction SP

Shiji Huang, Lei Ye, Min Chen, Wenhai Luo, Dihong Wang, Chenqi Xu, Deyuan Liang

A thorough understanding of the interaction between the target agent and surrounding agents is a prerequisite for accurate trajectory prediction. Although many methods have been explored, they all assign correlation coefficients to surrounding agents in a purely learning-based manner. In this study, we present ASPILin, which manually selects interacting agents and calculates their correlations instead of attention scores. Surprisingly, these simple modifications can significantly improve prediction performance and substantially reduce computational costs. Additionally, ASPILin models the interacting agents at each past time step separately, rather than only modeling the interacting agents at the current time step. This clarifies the causal chain of the target agent's historical trajectory and helps the model better understand dynamic interactions. We intentionally simplified our model in other aspects, such as map encoding. Remarkably, experiments conducted on the INTERACTION, highD, and CitySim datasets demonstrate that our method is efficient and straightforward, outperforming other state-of-the-art methods.

comment: code:https://github.com/kkk00714/ASPILin

♻ ☆ Conquering the Communication Constraints to Enable Large Pre-Trained Models in Federated Learning

Guangyu Sun, Umar Khalid, Matias Mendieta, Taojiannan Yang, Pu Wang, Minwoo Lee, Chen Chen

Federated learning (FL) has emerged as a promising paradigm for enabling the collaborative training of models without centralized access to the raw data on local devices. In the typical FL paradigm (e.g., FedAvg), model weights are sent to and from the server each round to participating clients. Recently, the use of small pre-trained models has been shown effective in federated learning optimization and improving convergence. However, recent state-of-the-art pre-trained models are getting more capable but also have more parameters. In conventional FL, sharing the enormous model weights can quickly put a massive communication burden on the system, especially if more capable models are employed. Can we find a solution to enable those strong and readily-available pre-trained models in FL to achieve excellent performance while simultaneously reducing the communication burden? To this end, we investigate the use of parameter-efficient fine-tuning in federated learning and thus introduce a new framework: FedPEFT. Specifically, we systemically evaluate the performance of FedPEFT across a variety of client stability, data distribution, and differential privacy settings. By only locally tuning and globally sharing a small portion of the model weights, significant reductions in the total communication overhead can be achieved while maintaining competitive or even better performance in a wide range of federated learning scenarios, providing insight into a new paradigm for practical and effective federated systems.

♻ ☆ Latent Noise Segmentation: How Neural Noise Leads to the Emergence of Segmentation and Grouping ICML 2024

Ben Lonnqvist, Zhengqing Wu, Michael H. Herzog

Humans are able to segment images effortlessly without supervision using perceptual grouping. Here, we propose a counter-intuitive computational approach to solving unsupervised perceptual grouping and segmentation: that they arise because of neural noise, rather than in spite of it. We (1) mathematically demonstrate that under realistic assumptions, neural noise can be used to separate objects from each other; (2) that adding noise in a DNN enables the network to segment images even though it was never trained on any segmentation labels; and (3) that segmenting objects using noise results in segmentation performance that aligns with the perceptual grouping phenomena observed in humans, and is sample-efficient. We introduce the Good Gestalt (GG) datasets -- six datasets designed to specifically test perceptual grouping, and show that our DNN models reproduce many important phenomena in human perception, such as illusory contours, closure, continuity, proximity, and occlusion. Finally, we (4) show that our model improves performance on our GG datasets compared to other tested unsupervised models by $24.9\%$. Together, our results suggest a novel unsupervised segmentation method requiring few assumptions, a new explanation for the formation of perceptual grouping, and a novel potential benefit of neural noise.

comment: ICML 2024 camera ready version

♻ ☆ STBA: Towards Evaluating the Robustness of DNNs for Query-Limited Black-box Scenario

Renyang Liu, Kwok-Yan Lam, Wei Zhou, Sixing Wu, Jun Zhao, Dongting Hu, Mingming Gong

Many attack techniques have been proposed to explore the vulnerability of DNNs and further help to improve their robustness. Despite the significant progress made recently, existing black-box attack methods still suffer from unsatisfactory performance due to the vast number of queries needed to optimize desired perturbations. Besides, the other critical challenge is that adversarial examples built in a noise-adding manner are abnormal and struggle to successfully attack robust models, whose robustness is enhanced by adversarial training against small perturbations. There is no doubt that these two issues mentioned above will significantly increase the risk of exposure and result in a failure to dig deeply into the vulnerability of DNNs. Hence, it is necessary to evaluate DNNs' fragility sufficiently under query-limited settings in a non-additional way. In this paper, we propose the Spatial Transform Black-box Attack (STBA), a novel framework to craft formidable adversarial examples in the query-limited scenario. Specifically, STBA introduces a flow field to the high-frequency part of clean images to generate adversarial examples and adopts the following two processes to enhance their naturalness and significantly improve the query efficiency: a) we apply an estimated flow field to the high-frequency part of clean images to generate adversarial examples instead of introducing external noise to the benign image, and b) we leverage an efficient gradient estimation method based on a batch of samples to optimize such an ideal flow field under query-limited settings. Compared to existing score-based black-box baselines, extensive experiments indicated that STBA could effectively improve the imperceptibility of the adversarial examples and remarkably boost the attack success rate under query-limited settings.

comment: Accepted by T-MM

♻ ☆ ODTFormer: Efficient Obstacle Detection and Tracking with Stereo Cameras Based on Transformer IROS 2024

Tianye Ding, Hongyu Li, Huaizu Jiang

Obstacle detection and tracking represent a critical component in robot autonomous navigation. In this paper, we propose ODTFormer, a Transformer-based model to address both obstacle detection and tracking problems. For the detection task, our approach leverages deformable attention to construct a 3D cost volume, which is decoded progressively in the form of voxel occupancy grids. We further track the obstacles by matching the voxels between consecutive frames. The entire model can be optimized in an end-to-end manner. Through extensive experiments on DrivingStereo and KITTI benchmarks, our model achieves state-of-the-art performance in the obstacle detection task. We also report comparable accuracy to state-of-the-art obstacle tracking models while requiring only a fraction of their computation cost, typically ten-fold to twenty-fold less. The code and model weights will be publicly released.

comment: 8 pages. Accepted by IROS 2024

♻ ☆ LLaVA-MoD: Making LLaVA Tiny via MoE Knowledge Distillation

Fangxun Shu, Yue Liao, Le Zhuo, Chenning Xu, Lei Zhang, Guanghao Zhang, Haonan Shi, Long Chen, Tao Zhong, Wanggui He, Siming Fu, Haoyuan Li, Bolin Li, Zhelun Yu, Si Liu, Hongsheng Li, Hao Jiang

We introduce LLaVA-MoD, a novel framework designed to enable the efficient training of small-scale Multimodal Language Models (s-MLLM) by distilling knowledge from large-scale MLLM (l-MLLM). Our approach tackles two fundamental challenges in MLLM distillation. First, we optimize the network structure of s-MLLM by integrating a sparse Mixture of Experts (MoE) architecture into the language model, striking a balance between computational efficiency and model expressiveness. Second, we propose a progressive knowledge transfer strategy to ensure comprehensive knowledge migration. This strategy begins with mimic distillation, where we minimize the Kullback-Leibler (KL) divergence between output distributions to enable the student model to emulate the teacher network's understanding. Following this, we introduce preference distillation via Direct Preference Optimization (DPO), where the key lies in treating l-MLLM as the reference model. During this phase, the s-MLLM's ability to discriminate between superior and inferior examples is significantly enhanced beyond l-MLLM, leading to a better student that surpasses its teacher, particularly in hallucination benchmarks. Extensive experiments demonstrate that LLaVA-MoD outperforms existing models across various multimodal benchmarks while maintaining a minimal number of activated parameters and low computational costs. Remarkably, LLaVA-MoD, with only 2B activated parameters, surpasses Qwen-VL-Chat-7B by an average of 8.8% across benchmarks, using merely 0.3% of the training data and 23% trainable parameters. These results underscore LLaVA-MoD's ability to effectively distill comprehensive knowledge from its teacher model, paving the way for the development of more efficient MLLMs. The code will be available on: https://github.com/shufangxun/LLaVA-MoD.

♻ ☆ DIP-Watermark: A Double Identity Protection Method Based on Robust Adversarial Watermark

Yunming Zhang, Dengpan Ye, Caiyun Xie, Sipeng Shen, Ziyi Liu, Jiacheng Deng, Long Tang

The wide deployment of Face Recognition (FR) systems poses privacy risks. One countermeasure is adversarial attack, deceiving unauthorized malicious FR, but it also disrupts regular identity verification of trusted authorizers, exacerbating the potential threat of identity impersonation. To address this, we propose the first double identity protection scheme based on traceable adversarial watermarking, termed DIP-Watermark. DIP-Watermark employs a one-time watermark embedding to deceive unauthorized FR models and allows authorizers to perform identity verification by extracting the watermark. Specifically, we propose an information-guided adversarial attack against FR models. The encoder embeds an identity-specific watermark into the deep feature space of the carrier, guiding recognizable features of the image to deviate from the source identity. We further adopt a collaborative meta-optimization strategy compatible with sub-tasks, which regularizes the joint optimization direction of the encoder and decoder. This strategy enhances the representation of universal carrier features, mitigating multi-objective optimization conflicts in watermarking. Experiments confirm that DIP-Watermark achieves significant attack success rates and traceability accuracy on state-of-the-art FR models, exhibiting remarkable robustness that outperforms the existing privacy protection methods using adversarial attacks and deep watermarking, or simple combinations of the two. Our work potentially opens up new insights into proactive protection for FR privacy.

♻ ☆ Hierarchical Light Transformer Ensembles for Multimodal Trajectory Forecasting

Adrien Lafage, Mathieu Barbier, Gianni Franchi, David Filliat

Accurate trajectory forecasting is crucial for the performance of various systems, such as advanced driver-assistance systems and self-driving vehicles. These forecasts allow to anticipate events leading to collisions and, therefore, to mitigate them. Deep Neural Networks have excelled in motion forecasting, but issues like overconfidence and uncertainty quantification persist. Deep Ensembles address these concerns, yet applying them to multimodal distributions remains challenging. In this paper, we propose a novel approach named Hierarchical Light Transformer Ensembles (HLT-Ens), aimed at efficiently training an ensemble of Transformer architectures using a novel hierarchical loss function. HLT-Ens leverages grouped fully connected layers, inspired by grouped convolution techniques, to capture multimodal distributions, effectively. Through extensive experimentation, we demonstrate that HLT-Ens achieves state-of-the-art performance levels, offering a promising avenue for improving trajectory forecasting techniques.

comment: acknowledgement added

♻ ☆ Pulling Target to Source: A New Perspective on Domain Adaptive Semantic Segmentation

Haochen Wang, Yujun Shen, Jingjing Fei, Wei Li, Liwei Wu, Yuxi Wang, Zhaoxiang Zhang

Domain adaptive semantic segmentation aims to transfer knowledge from a labeled source domain to an unlabeled target domain. However, existing methods primarily focus on directly learning qualified target features, making it challenging to guarantee their discrimination in the absence of target labels. This work provides a new perspective. We observe that the features learned with source data manage to keep categorically discriminative during training, thereby enabling us to implicitly learn adequate target representations by simply \textbf{pulling target features close to source features for each category}. To this end, we propose T2S-DA, which we interpret as a form of pulling Target to Source for Domain Adaptation, encouraging the model in learning similar cross-domain features. Also, considering the pixel categories are heavily imbalanced for segmentation datasets, we come up with a dynamic re-weighting strategy to help the model concentrate on those underperforming classes. Extensive experiments confirm that T2S-DA learns a more discriminative and generalizable representation, significantly surpassing the state-of-the-art. We further show that our method is quite qualified for the domain generalization task, verifying its domain-invariant property.

comment: Accepted by IJCV

♻ ☆ Improving Text Generation on Images with Synthetic Captions

Jun Young Koh, Sang Hyun Park, Joy Song

The recent emergence of latent diffusion models such as SDXL and SD 1.5 has shown significant capability in generating highly detailed and realistic images. Despite their remarkable ability to produce images, generating accurate text within images still remains a challenging task. In this paper, we examine the validity of fine-tuning approaches in generating legible text within the image. We propose a low-cost approach by leveraging SDXL without any time-consuming training on large-scale datasets. The proposed strategy employs a fine-tuning technique that examines the effects of data refinement levels and synthetic captions. Moreover, our results demonstrate how our small scale fine-tuning approach can improve the accuracy of text generation in different scenarios without the need of additional multimodal encoders. Our experiments show that with the addition of random letters to our raw dataset, our model's performance improves in producing well-formed visual text.

comment: 2024 16th IIAI International Congress on Advanced Applied Informatics (IIAI-AAI)

♻ ☆ Harmonizing Visual Text Comprehension and Generation NeurIPS 2024

Zhen Zhao, Jingqun Tang, Binghong Wu, Chunhui Lin, Shu Wei, Hao Liu, Xin Tan, Zhizhong Zhang, Can Huang, Yuan Xie

In this work, we present TextHarmony, a unified and versatile multimodal generative model proficient in comprehending and generating visual text. Simultaneously generating images and texts typically results in performance degradation due to the inherent inconsistency between vision and language modalities. To overcome this challenge, existing approaches resort to modality-specific data for supervised fine-tuning, necessitating distinct model instances. We propose Slide-LoRA, which dynamically aggregates modality-specific and modality-agnostic LoRA experts, partially decoupling the multimodal generation space. Slide-LoRA harmonizes the generation of vision and language within a singular model instance, thereby facilitating a more unified generative process. Additionally, we develop a high-quality image caption dataset, DetailedTextCaps-100K, synthesized with a sophisticated closed-source MLLM to enhance visual text generation capabilities further. Comprehensive experiments across various benchmarks demonstrate the effectiveness of the proposed approach. Empowered by Slide-LoRA, TextHarmony achieves comparable performance to modality-specific fine-tuning results with only a 2% increase in parameters and shows an average improvement of 2.5% in visual text comprehension tasks and 4.0% in visual text generation tasks. Our work delineates the viability of an integrated approach to multimodal generation within the visual text domain, setting a foundation for subsequent inquiries. Code is available at https://github.com/bytedance/TextHarmony.

comment: accepted by NeurIPS 2024

♻ ☆ Lightning-Fast Image Inversion and Editing for Text-to-Image Diffusion Models

Dvir Samuel, Barak Meiri, Haggai Maron, Yoad Tewel, Nir Darshan, Shai Avidan, Gal Chechik, Rami Ben-Ari

Diffusion inversion is the problem of taking an image and a text prompt that describes it and finding a noise latent that would generate the exact same image. Most current deterministic inversion techniques operate by approximately solving an implicit equation and may converge slowly or yield poor reconstructed images. We formulate the problem by finding the roots of an implicit equation and devlop a method to solve it efficiently. Our solution is based on Newton-Raphson (NR), a well-known technique in numerical analysis. We show that a vanilla application of NR is computationally infeasible while naively transforming it to a computationally tractable alternative tends to converge to out-of-distribution solutions, resulting in poor reconstruction and editing. We therefore derive an efficient guided formulation that fastly converges and provides high-quality reconstructions and editing. We showcase our method on real image editing with three popular open-sourced diffusion models: Stable Diffusion, SDXL-Turbo, and Flux with different deterministic schedulers. Our solution, Guided Newton-Raphson Inversion, inverts an image within 0.4 sec (on an A100 GPU) for few-step models (SDXL-Turbo and Flux.1), opening the door for interactive image editing. We further show improved results in image interpolation and generation of rare objects.

♻ ☆ Improving Instance Optimization in Deformable Image Registration with Gradient Projection MICCAI 2024

Yi Zhang, Yidong Zhao, Qian Tao

Deformable image registration is inherently a multi-objective optimization (MOO) problem, requiring a delicate balance between image similarity and deformation regularity. These conflicting objectives often lead to poor optimization outcomes, such as being trapped in unsatisfactory local minima or experiencing slow convergence. Deep learning methods have recently gained popularity in this domain due to their efficiency in processing large datasets and achieving high accuracy. However, they often underperform during test time compared to traditional optimization techniques, which further explore iterative, instance-specific gradient-based optimization. This performance gap is more pronounced when a distribution shift between training and test data exists. To address this issue, we focus on the instance optimization (IO) paradigm, which involves additional optimization for test-time instances based on a pre-trained model. IO effectively combines the generalization capabilities of deep learning with the fine-tuning advantages of instance-specific optimization. Within this framework, we emphasize the use of gradient projection to mitigate conflicting updates in MOO. This technique projects conflicting gradients into a common space, better aligning the dual objectives and enhancing optimization stability. We validate our method using a state-of-the-art foundation model on the 3D Brain inter-subject registration task (LUMIR) from the Learn2Reg 2024 Challenge. Our results show significant improvements over standard gradient descent, leading to more accurate and reliable registration results.

comment: Learn2Reg Challenge at MICCAI 2024

♻ ☆ ERX: A Fast Real-Time Anomaly Detection Algorithm for Hyperspectral Line Scanning

Samuel Garske, Bradley Evans, Christopher Artlett, KC Wong

Detecting unexpected objects (anomalies) in real time has great potential for monitoring, managing, and protecting the environment. Hyperspectral line-scan cameras are a low-cost solution that enhance confidence in anomaly detection over RGB and multispectral imagery. However, existing line-scan algorithms are too slow when using small computers (e.g. those onboard a drone or small satellite), do not adapt to changing scenery, or lack robustness against geometric distortions. This paper introduces the Exponentially moving RX algorithm (ERX) to address these issues, and compares it with existing RX-based anomaly detection methods for hyperspectral line scanning. Three large and more complex datasets are also introduced to better assess the practical challenges when using line-scan cameras (two hyperspectral and one multispectral). ERX is evaluated using a Jetson Xavier NX compute module, achieving the best combination of speed and detection performance. This research paves the way for future studies in grouping and locating anomalous objects, adaptive and automatic threshold selection, and real-time field tests. The datasets and the Python code are available at: https://github.com/WiseGamgee/HyperAD.

comment: 17 pages, 13 figures, 4 tables, code and datasets accessible at https://github.com/WiseGamgee/HyperAD

♻ ☆ CAT: Contrastive Adapter Training for Personalized Image Generation CVPR

Jae Wan Park, Sang Hyun Park, Jun Young Koh, Junha Lee, Min Song

The emergence of various adapters, including Low-Rank Adaptation (LoRA) applied from the field of natural language processing, has allowed diffusion models to personalize image generation at a low cost. However, due to the various challenges including limited datasets and shortage of regularization and computation resources, adapter training often results in unsatisfactory outcomes, leading to the corruption of the backbone model's prior knowledge. One of the well known phenomena is the loss of diversity in object generation, especially within the same class which leads to generating almost identical objects with minor variations. This poses challenges in generation capabilities. To solve this issue, we present Contrastive Adapter Training (CAT), a simple yet effective strategy to enhance adapter training through the application of CAT loss. Our approach facilitates the preservation of the base model's original knowledge when the model initiates adapters. Furthermore, we introduce the Knowledge Preservation Score (KPS) to evaluate CAT's ability to keep the former information. We qualitatively and quantitatively compare CAT's improvement. Finally, we mention the possibility of CAT in the aspects of multi-concept adapter and optimization.

comment: CVPRW 2024

♻ ☆ Gaussian-Informed Continuum for Physical Property Identification and Simulation NeurIPS 2024

Junhao Cai, Yuji Yang, Weihao Yuan, Yisheng He, Zilong Dong, Liefeng Bo, Hui Cheng, Qifeng Chen

This paper studies the problem of estimating physical properties (system identification) through visual observations. To facilitate geometry-aware guidance in physical property estimation, we introduce a novel hybrid framework that leverages 3D Gaussian representation to not only capture explicit shapes but also enable the simulated continuum to render object masks as 2D shape surrogates during training. We propose a new dynamic 3D Gaussian framework based on motion factorization to recover the object as 3D Gaussian point sets across different time states. Furthermore, we develop a coarse-to-fine filling strategy to generate the density fields of the object from the Gaussian reconstruction, allowing for the extraction of object continuums along with their surfaces and the integration of Gaussian attributes into these continuums. In addition to the extracted object surfaces, the Gaussian-informed continuum also enables the rendering of object masks during simulations, serving as 2D-shape guidance for physical property estimation. Extensive experimental evaluations demonstrate that our pipeline achieves state-of-the-art performance across multiple benchmarks and metrics. Additionally, we illustrate the effectiveness of the proposed method through real-world demonstrations, showcasing its practical utility. Our project page is at https://jukgei.github.io/project/gic.

comment: 21 pages, 8 figures, NeurIPS 2024

♻ ☆ LVBench: An Extreme Long Video Understanding Benchmark

Weihan Wang, Zehai He, Wenyi Hong, Yean Cheng, Xiaohan Zhang, Ji Qi, Xiaotao Gu, Shiyu Huang, Bin Xu, Yuxiao Dong, Ming Ding, Jie Tang

Recent progress in multimodal large language models has markedly enhanced the understanding of short videos (typically under one minute), and several evaluation datasets have emerged accordingly. However, these advancements fall short of meeting the demands of real-world applications such as embodied intelligence for long-term decision-making, in-depth movie reviews and discussions, and live sports commentary, all of which require comprehension of long videos spanning several hours. To address this gap, we introduce LVBench, a benchmark specifically designed for long video understanding. Our dataset comprises publicly sourced videos and encompasses a diverse set of tasks aimed at long video comprehension and information extraction. LVBench is designed to challenge multimodal models to demonstrate long-term memory and extended comprehension capabilities. Our extensive evaluations reveal that current multimodal models still underperform on these demanding long video understanding tasks. Through LVBench, we aim to spur the development of more advanced models capable of tackling the complexities of long video comprehension. Our data and code are publicly available at: https://lvbench.github.io.

♻ ☆ Advancing Open-Set Domain Generalization Using Evidential Bi-Level Hardest Domain Scheduler NeurIPS 2024

Kunyu Peng, Di Wen, Kailun Yang, Ao Luo, Yufan Chen, Jia Fu, M. Saquib Sarfraz, Alina Roitberg, Rainer Stiefelhagen

In Open-Set Domain Generalization (OSDG), the model is exposed to both new variations of data appearance (domains) and open-set conditions, where both known and novel categories are present at test time. The challenges of this task arise from the dual need to generalize across diverse domains and accurately quantify category novelty, which is critical for applications in dynamic environments. Recently, meta-learning techniques have demonstrated superior results in OSDG, effectively orchestrating the meta-train and -test tasks by employing varied random categories and predefined domain partition strategies. These approaches prioritize a well-designed training schedule over traditional methods that focus primarily on data augmentation and the enhancement of discriminative feature learning. The prevailing meta-learning models in OSDG typically utilize a predefined sequential domain scheduler to structure data partitions. However, a crucial aspect that remains inadequately explored is the influence brought by strategies of domain schedulers during training. In this paper, we observe that an adaptive domain scheduler benefits more in OSDG compared with prefixed sequential and random domain schedulers. We propose the Evidential Bi-Level Hardest Domain Scheduler (EBiL-HaDS) to achieve an adaptive domain scheduler. This method strategically sequences domains by assessing their reliabilities in utilizing a follower network, trained with confidence scores learned in an evidential manner, regularized by max rebiasing discrepancy, and optimized in a bi-level manner. The results show that our method substantially improves OSDG performance and achieves more discriminative embeddings for both the seen and unseen categories. The source code is publicly available at https://github.com/KPeng9510/EBiL-HaDS.

comment: Accepted to NeurIPS 2024. The source code is publicly available at https://github.com/KPeng9510/EBiL-HaDS

♻ ☆ Learning to Manipulate Anywhere: A Visual Generalizable Framework For Reinforcement Learning

Zhecheng Yuan, Tianming Wei, Shuiqi Cheng, Gu Zhang, Yuanpei Chen, Huazhe Xu

Can we endow visuomotor robots with generalization capabilities to operate in diverse open-world scenarios? In this paper, we propose \textbf{Maniwhere}, a generalizable framework tailored for visual reinforcement learning, enabling the trained robot policies to generalize across a combination of multiple visual disturbance types. Specifically, we introduce a multi-view representation learning approach fused with Spatial Transformer Network (STN) module to capture shared semantic information and correspondences among different viewpoints. In addition, we employ a curriculum-based randomization and augmentation approach to stabilize the RL training process and strengthen the visual generalization ability. To exhibit the effectiveness of Maniwhere, we meticulously design 8 tasks encompassing articulate objects, bi-manual, and dexterous hand manipulation tasks, demonstrating Maniwhere's strong visual generalization and sim2real transfer abilities across 3 hardware platforms. Our experiments show that Maniwhere significantly outperforms existing state-of-the-art methods. Videos are provided at https://gemcollector.github.io/maniwhere/.

comment: Webpage: https://gemcollector.github.io/maniwhere/

♻ ☆ The Ultimate Combo: Boosting Adversarial Example Transferability by Composing Data Augmentations

Zebin Yun, Achi-Or Weingarten, Eyal Ronen, Mahmood Sharif

To help adversarial examples generalize from surrogate machine-learning (ML) models to targets, certain transferability-based black-box evasion attacks incorporate data augmentations (e.g., random resizing). Yet, prior work has explored limited augmentations and their composition. To fill the gap, we systematically studied how data augmentation affects transferability. Specifically, we explored 46 augmentation techniques originally proposed to help ML models generalize to unseen benign samples, and assessed how they impact transferability, when applied individually or composed. Performing exhaustive search on a small subset of augmentation techniques and genetic search on all techniques, we identified augmentation combinations that help promote transferability. Extensive experiments with the ImageNet and CIFAR-10 datasets and 18 models showed that simple color-space augmentations (e.g., color to greyscale) attain high transferability when combined with standard augmentations. Furthermore, we discovered that composing augmentations impacts transferability mostly monotonically (i.e., more augmentations $\rightarrow$ $\ge$transferability). We also found that the best composition significantly outperformed the state of the art (e.g., 91.8% vs. $\le$82.5% average transferability to adversarially trained targets on ImageNet). Lastly, our theoretical analysis, backed by empirical evidence, intuitively explains why certain augmentations promote transferability.

comment: Accepted by AISec'24

♻ ☆ Diffusion Models are Certifiably Robust Classifiers NeurIPS 2024

Huanran Chen, Yinpeng Dong, Shitong Shao, Zhongkai Hao, Xiao Yang, Hang Su, Jun Zhu

Generative learning, recognized for its effective modeling of data distributions, offers inherent advantages in handling out-of-distribution instances, especially for enhancing robustness to adversarial attacks. Among these, diffusion classifiers, utilizing powerful diffusion models, have demonstrated superior empirical robustness. However, a comprehensive theoretical understanding of their robustness is still lacking, raising concerns about their vulnerability to stronger future attacks. In this study, we prove that diffusion classifiers possess $O(1)$ Lipschitzness, and establish their certified robustness, demonstrating their inherent resilience. To achieve non-constant Lipschitzness, thereby obtaining much tighter certified robustness, we generalize diffusion classifiers to classify Gaussian-corrupted data. This involves deriving the evidence lower bounds (ELBOs) for these distributions, approximating the likelihood using the ELBO, and calculating classification probabilities via Bayes' theorem. Experimental results show the superior certified robustness of these Noised Diffusion Classifiers (NDCs). Notably, we achieve over 80% and 70% certified robustness on CIFAR-10 under adversarial perturbations with \(\ell_2\) norms less than 0.25 and 0.5, respectively, using a single off-the-shelf diffusion model without any additional data.

comment: Accepted by NeurIPS 2024

♻ ☆ SemiSAM: Enhancing Semi-Supervised Medical Image Segmentation via SAM-Assisted Consistency Regularization

Yichi Zhang, Jin Yang, Yuchen Liu, Yuan Cheng, Yuan Qi

Semi-supervised learning has attracted much attention due to its less dependence on acquiring abundant annotations from experts compared to fully supervised methods, which is especially important for medical image segmentation which typically requires intensive pixel/voxel-wise labeling by domain experts. Although semi-supervised methods can improve the performance by utilizing unlabeled data, there are still gaps between fully supervised methods under extremely limited annotation scenarios. In this paper, we propose a simple yet efficient strategy to explore the usage of the Segment Anything Model (SAM) for enhancing semi-supervised medical image segmentation. Concretely, the segmentation model trained with domain knowledge provides information for localization and generating input prompts to the SAM. Then the generated pseudo-labels of SAM are utilized as additional supervision to assist in the learning procedure of the semi-supervised framework. Extensive experiments demonstrate that SemiSAM significantly improves the performance of existing semi-supervised frameworks when only one or a few labeled images are available and shows strong efficiency as a plug-and-play strategy for semi-supervised medical image segmentation.

comment: Accept for BIBM 2024

♻ ☆ RotCAtt-TransUNet++: Novel Deep Neural Network for Sophisticated Cardiac Segmentation

Quoc-Bao Nguyen-Le, Tuan-Hy Le, Anh-Triet Do, Quoc-Huy Trinh

Cardiovascular disease remains a predominant global health concern, responsible for a significant portion of mortality worldwide. Accurate segmentation of cardiac medical imaging data is pivotal in mitigating fatality rates associated with cardiovascular conditions. However, existing state-of-the-art (SOTA) neural networks, including both CNN-based and Transformer-based approaches, exhibit limitations in practical applicability due to their inability to effectively capture inter-slice connections alongside intra-slice information. This deficiency is particularly evident in datasets featuring intricate, long-range details along the z-axis, such as coronary arteries in axial views. Additionally, SOTA methods fail to differentiate non-cardiac components from myocardium in segmentation, leading to the "spraying" phenomenon. To address these challenges, we present RotCAtt-TransUNet++, a novel architecture tailored for robust segmentation of complex cardiac structures. Our approach emphasizes modeling global contexts by aggregating multiscale features with nested skip connections in the encoder. It integrates transformer layers to capture interactions between patches and employs a rotatory attention mechanism to capture connectivity between multiple slices (inter-slice information). Additionally, a channel-wise cross-attention gate guides the fused multi-scale channel-wise information and features from decoder stages to bridge semantic gaps. Experimental results demonstrate that our proposed model outperforms existing SOTA approaches across four cardiac datasets and one abdominal dataset. Importantly, coronary arteries and myocardium are annotated with near-perfect accuracy during inference. An ablation study shows that the rotatory attention mechanism effectively transforms embedded vectorized patches in the semantic dimensional space, enhancing segmentation accuracy.

comment: 11 pages, 11 figures

♻ ☆ Exploring Self-Supervised Skeleton-Based Human Action Recognition under Occlusions

Yifei Chen, Kunyu Peng, Alina Roitberg, David Schneider, Jiaming Zhang, Junwei Zheng, Ruiping Liu, Yufan Chen, Kailun Yang, Rainer Stiefelhagen

To integrate self-supervised skeleton-based action recognition methods into autonomous robotic systems, it is crucial to consider adverse situations involving target occlusions. Such a scenario, despite its practical relevance, is rarely addressed in existing self-supervised skeleton-based action recognition methods. To empower models with the capacity to address occlusion, we propose a simple and effective method. We first pre-train using occluded skeleton sequences, then use k-means clustering (KMeans) on sequence embeddings to group semantically similar samples. Next, we propose KNN-Imputation to fill in missing skeleton data based on the closest sample neighbors. Imputing incomplete skeleton sequences to create relatively complete sequences as input provides significant benefits to existing skeleton-based self-supervised methods. Meanwhile, building on the state-of-the-art Partial Spatio-Temporal Learning (PSTL), we introduce an Occluded Partial Spatio-Temporal Learning (OPSTL) framework. This enhancement utilizes Adaptive Spatial Masking (ASM) for better use of high-quality, intact skeletons. The new proposed method is verified on the challenging occluded versions of the NTURGB+D 60 and NTURGB+D 120. The source code is publicly available at https://github.com/cyfml/OPSTL.

comment: The source code is publicly available at https://github.com/cyfml/OPSTL

♻ ☆ RealignDiff: Boosting Text-to-Image Diffusion Model with Coarse-to-fine Semantic Re-alignment

Guian Fang, Zutao Jiang, Jianhua Han, Guansong Lu, Hang Xu, Shengcai Liao, Xiaojun Chang, Xiaodan Liang

Recent advances in text-to-image diffusion models have achieved remarkable success in generating high-quality, realistic images from textual descriptions. However, these approaches have faced challenges in precisely aligning the generated visual content with the textual concepts described in the prompts. In this paper, we propose a two-stage coarse-to-fine semantic re-alignment method, named RealignDiff, aimed at improving the alignment between text and images in text-to-image diffusion models. In the coarse semantic re-alignment phase, a novel caption reward, leveraging the BLIP-2 model, is proposed to evaluate the semantic discrepancy between the generated image caption and the given text prompt. Subsequently, the fine semantic re-alignment stage employs a local dense caption generation module and a re-weighting attention modulation module to refine the previously generated images from a local semantic view. Experimental results on the MS-COCO and ViLG-300 datasets demonstrate that the proposed two-stage coarse-to-fine semantic re-alignment method outperforms other baseline re-alignment techniques by a substantial margin in both visual quality and semantic similarity with the input prompt.

♻ ☆ Real-World Robot Applications of Foundation Models: A Review

Kento Kawaharazuka, Tatsuya Matsushima, Andrew Gambardella, Jiaxian Guo, Chris Paxton, Andy Zeng

Recent developments in foundation models, like Large Language Models (LLMs) and Vision-Language Models (VLMs), trained on extensive data, facilitate flexible application across different tasks and modalities. Their impact spans various fields, including healthcare, education, and robotics. This paper provides an overview of the practical application of foundation models in real-world robotics, with a primary emphasis on the replacement of specific components within existing robot systems. The summary encompasses the perspective of input-output relationships in foundation models, as well as their role in perception, motion planning, and control within the field of robotics. This paper concludes with a discussion of future challenges and implications for practical robot applications.

♻ ☆ MMBench-Video: A Long-Form Multi-Shot Benchmark for Holistic Video Understanding NeurIPS 2024

Xinyu Fang, Kangrui Mao, Haodong Duan, Xiangyu Zhao, Yining Li, Dahua Lin, Kai Chen

The advent of large vision-language models (LVLMs) has spurred research into their applications in multi-modal contexts, particularly in video understanding. Traditional VideoQA benchmarks, despite providing quantitative metrics, often fail to encompass the full spectrum of video content and inadequately assess models' temporal comprehension. To address these limitations, we introduce MMBench-Video, a quantitative benchmark designed to rigorously evaluate LVLMs' proficiency in video understanding. MMBench-Video incorporates lengthy videos from YouTube and employs free-form questions, mirroring practical use cases. The benchmark is meticulously crafted to probe the models' temporal reasoning skills, with all questions human-annotated according to a carefully constructed ability taxonomy. We employ GPT-4 for automated assessment, demonstrating superior accuracy and robustness over earlier LLM-based evaluations. Utilizing MMBench-Video, we have conducted comprehensive evaluations that include both proprietary and open-source LVLMs for images and videos. MMBench-Video stands as a valuable resource for the research community, facilitating improved evaluation of LVLMs and catalyzing progress in the field of video understanding. The evalutation code of MMBench-Video will be integrated into VLMEvalKit: https://github.com/open-compass/VLMEvalKit.

comment: Accepted in NeurIPS 2024 Datasets and Benchmarks Track

♻ ☆ Selective Vision is the Challenge for Visual Reasoning: A Benchmark for Visual Argument Understanding EMNLP 2024

Jiwan Chung, Sungjae Lee, Minseo Kim, Seungju Han, Ashkan Yousefpour, Jack Hessel, Youngjae Yu

Visual arguments, often used in advertising or social causes, rely on images to persuade viewers to do or believe something. Understanding these arguments requires selective vision: only specific visual stimuli within an image are relevant to the argument, and relevance can only be understood within the context of a broader argumentative structure. While visual arguments are readily appreciated by human audiences, we ask: are today's AI capable of similar understanding? We present VisArgs, a dataset of 1,611 images annotated with 5,112 visual premises (with regions), 5,574 commonsense premises, and reasoning trees connecting them into structured arguments. We propose three tasks for evaluating visual argument understanding: premise localization, premise identification, and conclusion deduction. Experiments show that 1) machines struggle to capture visual cues: GPT-4-O achieved 78.5% accuracy, while humans reached 98.0%. Models also performed 19.5% worse when distinguishing between irrelevant objects within the image compared to external objects. 2) Providing relevant visual premises improved model performance significantly.

comment: 12 pages, 6 figures. Accepted as main paper in EMNLP 2024

♻ ☆ Can visual language models resolve textual ambiguity with visual cues? Let visual puns tell you! EMNLP 2024

Jiwan Chung, Seungwon Lim, Jaehyun Jeon, Seungbeen Lee, Youngjae Yu

Humans possess multimodal literacy, allowing them to actively integrate information from various modalities to form reasoning. Faced with challenges like lexical ambiguity in text, we supplement this with other modalities, such as thumbnail images or textbook illustrations. Is it possible for machines to achieve a similar multimodal understanding capability? In response, we present Understanding Pun with Image Explanations (UNPIE), a novel benchmark designed to assess the impact of multimodal inputs in resolving lexical ambiguities. Puns serve as the ideal subject for this evaluation due to their intrinsic ambiguity. Our dataset includes 1,000 puns, each accompanied by an image that explains both meanings. We pose three multimodal challenges with the annotations to assess different aspects of multimodal literacy; Pun Grounding, Disambiguation, and Reconstruction. The results indicate that various Socratic Models and Visual-Language Models improve over the text-only models when given visual context, particularly as the complexity of the tasks increases.

comment: Accepted as main paper in EMNLP 2024

♻ ☆ CV-VAE: A Compatible Video VAE for Latent Generative Video Models

Sijie Zhao, Yong Zhang, Xiaodong Cun, Shaoshu Yang, Muyao Niu, Xiaoyu Li, Wenbo Hu, Ying Shan

Spatio-temporal compression of videos, utilizing networks such as Variational Autoencoders (VAE), plays a crucial role in OpenAI's SORA and numerous other video generative models. For instance, many LLM-like video models learn the distribution of discrete tokens derived from 3D VAEs within the VQVAE framework, while most diffusion-based video models capture the distribution of continuous latent extracted by 2D VAEs without quantization. The temporal compression is simply realized by uniform frame sampling which results in unsmooth motion between consecutive frames. Currently, there lacks of a commonly used continuous video (3D) VAE for latent diffusion-based video models in the research community. Moreover, since current diffusion-based approaches are often implemented using pre-trained text-to-image (T2I) models, directly training a video VAE without considering the compatibility with existing T2I models will result in a latent space gap between them, which will take huge computational resources for training to bridge the gap even with the T2I models as initialization. To address this issue, we propose a method for training a video VAE of latent video models, namely CV-VAE, whose latent space is compatible with that of a given image VAE, e.g., image VAE of Stable Diffusion (SD). The compatibility is achieved by the proposed novel latent space regularization, which involves formulating a regularization loss using the image VAE. Benefiting from the latent space compatibility, video models can be trained seamlessly from pre-trained T2I or video models in a truly spatio-temporally compressed latent space, rather than simply sampling video frames at equal intervals. With our CV-VAE, existing video models can generate four times more frames with minimal finetuning. Extensive experiments are conducted to demonstrate the effectiveness of the proposed video VAE.

comment: Project Page: https://ailab-cvc.github.io/cvvae/index.html

♻ ☆ GPHM: Gaussian Parametric Head Model for Monocular Head Avatar Reconstruction

Yuelang Xu, Zhaoqi Su, Qingyao Wu, Yebin Liu

Creating high-fidelity 3D human head avatars is crucial for applications in VR/AR, digital human, and film production. Recent advances have leveraged morphable face models to generate animated head avatars from easily accessible data, representing varying identities and expressions within a low-dimensional parametric space. However, existing methods often struggle with modeling complex appearance details, e.g., hairstyles, and suffer from low rendering quality and efficiency. In this paper we introduce a novel approach, 3D Gaussian Parametric Head Model, which employs 3D Gaussians to accurately represent the complexities of the human head, allowing precise control over both identity and expression. The Gaussian model can handle intricate details, enabling realistic representations of varying appearances and complex expressions. Furthermore, we presents a well-designed training framework to ensure smooth convergence, providing a robust guarantee for learning the rich content. Our method achieves high-quality, photo-realistic rendering with real-time efficiency, making it a valuable contribution to the field of parametric head models. Finally, we apply the 3D Gaussian Parametric Head Model to monocular video or few-shot head avatar reconstruction tasks, which enables instant reconstruction of high-quality 3D head avatars even when input data is extremely limited, surpassing previous methods in terms of reconstruction quality and training speed.

comment: Project page: https://yuelangx.github.io/gphmv2/

♻ ☆ CoIN: A Benchmark of Continual Instruction tuNing for Multimodel Large Language Model

Cheng Chen, Junchen Zhu, Xu Luo, Hengtao Shen, Lianli Gao, Jingkuan Song

Instruction tuning represents a prevalent strategy employed by Multimodal Large Language Models (MLLMs) to align with human instructions and adapt to new tasks. Nevertheless, MLLMs encounter the challenge of adapting to users' evolving knowledge and demands. Therefore, how to retain existing skills while acquiring new knowledge needs to be investigated. In this paper, we present a comprehensive benchmark, namely Continual Instruction tuNing (CoIN), to assess existing MLLMs in the sequential instruction tuning paradigm. CoIN comprises 10 commonly used datasets spanning 8 task categories, ensuring a diverse range of instructions and tasks. Besides, the trained model is evaluated from two aspects: Instruction Following and General Knowledge, which assess the alignment with human intention and knowledge preserved for reasoning, respectively. Experiments on CoIN demonstrate that current powerful MLLMs still suffer catastrophic forgetting, and the failure in intention alignment assumes the main responsibility, instead of the knowledge forgetting. To this end, we introduce MoELoRA to MLLMs which is effective to retain the previous instruction alignment. Experimental results consistently illustrate the forgetting decreased from this method on CoIN.

♻ ☆ CD-NGP: A Fast Scalable Continual Representation for Dynamic Scenes

Zhenhuan Liu, Shuai Liu, Zhiwei Ning, Jie Yang, Wei Liu

Current methodologies for novel view synthesis (NVS) in dynamic scenes encounter significant challenges in harmonizing memory consumption, model complexity, training efficiency, and rendering fidelity. Existing offline techniques, while delivering high-quality results, are often characterized by substantial memory demands and limited scalability. In contrast, online methods grapple with the challenge of balancing rapid convergence with model compactness. To address these issues, we propose continual dynamic neural graphics primitives (CD-NGP). Our approach synergizes features from both temporal and spatial hash encodings to achieve high rendering quality, employs parameter reuse to enhance scalability, and leverages a continual learning framework to mitigate memory overhead. Furthermore, we introduce a novel dataset comprising multi-view, exceptionally long video sequences with substantial rigid and non-rigid motion, thereby substantiating the scalability of our method.

comment: new template, editing

♻ ☆ Hybrid Spatial Representations for Species Distribution Modeling SDM

Shiran Yuan, Hao Zhao

We address an important problem in ecology called Species Distribution Modeling (SDM), whose goal is to predict whether a species exists at a certain position on Earth. In particular, we tackle a challenging version of this task, where we learn from presence-only data in a community-sourced dataset, model a large number of species simultaneously, and do not use any additional environmental information. Previous work has used neural implicit representations to construct models that achieve promising results. However, implicit representations often generate predictions of limited spatial precision. We attribute this limitation to their inherently global formulation and inability to effectively capture local feature variations. This issue is especially pronounced with presence-only data and a large number of species. To address this, we propose a hybrid embedding scheme that combines both implicit and explicit embeddings. Specifically, the explicit embedding is implemented with a multiresolution hashgrid, enabling our models to better capture local information. Experiments demonstrate that our results exceed other works by a large margin on various standard benchmarks, and that the hybrid representation is better than both purely implicit and explicit ones. Qualitative visualizations and comprehensive ablation studies reveal that our hybrid representation successfully addresses the two main challenges. Our code is open-sourced at https://github.com/Shiran-Yuan/HSR-SDM.

comment: Project codebase https://github.com/Shiran-Yuan/HSR-SDM

☆ B-Side: Binary-Level Static System Call Identification

Gaspard Thévenon, Kevin Nguetchouang, Kahina Lazri, Alain Tchana, Pierre Olivier

System call filtering is widely used to secure programs in multi-tenant environments, and to sandbox applications in modern desktop software deployment and package management systems. Filtering rules are hard to write and maintain manually, hence generating them automatically is essential. To that aim, analysis tools able to identify every system call that can legitimately be invoked by a program are needed. Existing static analysis works lack precision because of a high number of false positives, and/or assume the availability of program/libraries source code -- something unrealistic in many scenarios such as cloud production environments. We present B-Side, a static binary analysis tool able to identify a superset of the system calls that an x86-64 static/dynamic executable may invoke at runtime. B-Side assumes no access to program/libraries sources, and shows a good degree of precision by leveraging symbolic execution, combined with a heuristic to detect system call wrappers, which represent an important source of precision loss in existing works. B-Side also allows to statically detect phases of execution in a program in which different filtering rules can be applied. We validate B-Side and demonstrate its higher precision compared to state-of-the-art works: over a set of popular applications, B-Side's average $F_1$ score is 0.81, vs. 0.31 and 0.53 for competitors. Over 557 static and dynamically-compiled binaries taken from the Debian repositories, B-Side identifies an average of 43 system calls, vs. 271 and 95 for two state-of-the art competitors. We further evaluate the strictness of the phase-based filtering policies that can be obtained with B-Side.

comment: Accepted to appear in the 25th ACM/IFIP International Middleware Conference (Middleware'24)

☆ Federated Transformer: Multi-Party Vertical Federated Learning on Practical Fuzzily Linked Data

Zhaomin Wu, Junyi Hou, Yiqun Diao, Bingsheng He

Federated Learning (FL) is an evolving paradigm that enables multiple parties to collaboratively train models without sharing raw data. Among its variants, Vertical Federated Learning (VFL) is particularly relevant in real-world, cross-organizational collaborations, where distinct features of a shared instance group are contributed by different parties. In these scenarios, parties are often linked using fuzzy identifiers, leading to a common practice termed as multi-party fuzzy VFL. Existing models generally address either multi-party VFL or fuzzy VFL between two parties. Extending these models to practical multi-party fuzzy VFL typically results in significant performance degradation and increased costs for maintaining privacy. To overcome these limitations, we introduce the Federated Transformer (FeT), a novel framework that supports multi-party VFL with fuzzy identifiers. FeT innovatively encodes these identifiers into data representations and employs a transformer architecture distributed across different parties, incorporating three new techniques to enhance performance. Furthermore, we have developed a multi-party privacy framework for VFL that integrates differential privacy with secure multi-party computation, effectively protecting local representations while minimizing associated utility costs. Our experiments demonstrate that the FeT surpasses the baseline models by up to 46\% in terms of accuracy when scaled to 50 parties. Additionally, in two-party fuzzy VFL settings, FeT also shows improved performance and privacy over cutting-edge VFL models.

☆ Multi-Continental Healthcare Modelling Using Blockchain-Enabled Federated Learning

Rui Sun, Zhipeng Wang, Hengrui Zhang, Ming Jiang, Yizhe Wen, Jiqun Zhang, Jiahao Sun, Shuoying Zhang, Erwu Liu, Kezhi Li

One of the biggest challenges of building artificial intelligence (AI) model in healthcare area is the data sharing. Since healthcare data is private, sensitive, and heterogeneous, collecting sufficient data for modelling is exhausted, costly, and sometimes impossible. In this paper, we propose a framework for global healthcare modelling using datasets from multi-continents (Europe, North America and Asia) while without sharing the local datasets, and choose glucose management as a study model to verify its effectiveness. Technically, blockchain-enabled federated learning is implemented with adaption to make it meet with the privacy and safety requirements of healthcare data, meanwhile rewards honest participation and penalize malicious activities using its on-chain incentive mechanism. Experimental results show that the proposed framework is effective, efficient, and privacy preserved. Its prediction accuracy is much better than the models trained from limited personal data and is similar to, and even slightly better than, the results from a centralized dataset. This work paves the way for international collaborations on healthcare projects, where additional data is crucial for reducing bias and providing benefits to humanity.

comment: Accepted by IEEE Global Blockchain Conference

☆ SJMalloc: the security-conscious, fast, thread-safe and memory-efficient heap allocator

Stephan Bauroth

Heap-based exploits that leverage memory management errors continue to pose a significant threat to application security. The root cause of these vulnerabilities are the memory management errors within the applications, however various hardened allocator designs have been proposed as mitigation. A common feature of these designs is the strategic decision to store heap metadata separately from the application data in use, thereby reducing the risk of metadata corruption leading to security breaches. Despite their potential benefits, hardened allocators have not been widely adopted in real-world applications. The primary barrier to their adoption is the performance overheads they introduce. These overheads can negatively impact the efficiency and speed of applications, which is a critical consideration for developers and system administrators. Having learned from previous implementations, we developed SJMalloc, a general-purpose, high-performance allocator that addresses these concerns. SJMalloc stores its metadata out-of-band, away from the application's data on the heap. This design choice not only enhances security but also improves performance. Across a variety of real-world workloads, SJMalloc demonstrates a ~6% performance improvement compared to GLibcs allocator, while using only ~5% more memory. Furthermore, SJMalloc successfully passes the generic elements of the GLibc malloc testsuite and can thus be used as a drop-in replacement for the standard allocator, offering an easy upgrade path for enhanced security and performance without requiring changes to existing applications.

☆ Securing Stack Smashing Protection in WebAssembly Applications CCS

Quentin Michaud, Yohan Pipereau, Olivier Levillain, Dhouha Ayed

WebAssembly is an instruction set architecture and binary format standard, designed for secure execution by an interpreter. Previous work has shown that WebAssembly is vulnerable to buffer overflow due to the lack of effective protection mechanisms. In this paper, we evaluate the implementation of Stack Smashing Protection (SSP) in WebAssembly standalone runtimes, and uncover two weaknesses in their current implementation. The first one is the possibility to overwrite the SSP reference value because of the contiguous memory zones inside a WebAssembly process. The second comes from the reliance of WebAssembly on the runtime to provide randomness in order to initialize the SSP reference value, which impacts the robustness of the solution. We address these two flaws by hardening the SSP implementation in terms of storage and random generator failure, in a way that is generalizable to all of WebAssembly. We evaluate our new, more robust, solution to prove that the implemented improvements do not reduce the efficiency of SSP.

comment: Presented at PLAS 24 (ACM CCS workshop) https://plas24.github.io/

☆ Slot: Provenance-Driven APT Detection through Graph Reinforcement Learning

Wei Qiao, Yebo Feng, Teng Li, Zijian Zhang, Zhengzi Xu, Zhuo Ma, Yulong Shen, JianFeng Ma, Yang Liu

Advanced Persistent Threats (APTs) represent sophisticated cyberattacks characterized by their ability to remain undetected within the victim system for extended periods, aiming to exfiltrate sensitive data or disrupt operations. Existing detection approaches often struggle to effectively identify these complex threats, construct the attack chain for defense facilitation, or resist adversarial attacks. To overcome these challenges, we propose Slot, an advanced APT detection approach based on provenance graphs and graph reinforcement learning. Slot excels in uncovering multi-level hidden relationships, such as causal, contextual, and indirect connections, among system behaviors through provenance graph mining. By pioneering the integration of graph reinforcement learning, Slot dynamically adapts to new user activities and evolving attack strategies, enhancing its resilience against adversarial attacks. Additionally, Slot automatically constructs the attack chain according to detected attacks with clustering algorithms, providing precise identification of attack paths and facilitating the development of defense strategies. Evaluations with real-world datasets demonstrate Slot's outstanding accuracy, efficiency, adaptability, and robustness in APT detection, with most metrics surpassing state-of-the-art methods. Additionally, case studies conducted to assess Slot's effectiveness in supporting APT defense further establish it as a practical and reliable tool for cybersecurity protection.

☆ Enhancing Federated Learning Convergence with Dynamic Data Queue and Data Entropy-driven Participant Selection

Charuka Herath, Xiaolan Liu, Sangarapillai Lambotharan, Yogachandran Rahulamathavan

Federated Learning (FL) is a decentralized approach for collaborative model training on edge devices. This distributed method of model training offers advantages in privacy, security, regulatory compliance, and cost-efficiency. Our emphasis in this research lies in addressing statistical complexity in FL, especially when the data stored locally across devices is not identically and independently distributed (non-IID). We have observed an accuracy reduction of up to approximately 10\% to 30\%, particularly in skewed scenarios where each edge device trains with only 1 class of data. This reduction is attributed to weight divergence, quantified using the Euclidean distance between device-level class distributions and the population distribution, resulting in a bias term (\(\delta_k\)). As a solution, we present a method to improve convergence in FL by creating a global subset of data on the server and dynamically distributing it across devices using a Dynamic Data queue-driven Federated Learning (DDFL). Next, we leverage Data Entropy metrics to observe the process during each training round and enable reasonable device selection for aggregation. Furthermore, we provide a convergence analysis of our proposed DDFL to justify their viability in practical FL scenarios, aiming for better device selection, a non-sub-optimal global model, and faster convergence. We observe that our approach results in a substantial accuracy boost of approximately 5\% for the MNIST dataset, around 18\% for CIFAR-10, and 20\% for CIFAR-100 with a 10\% global subset of data, outperforming the state-of-the-art (SOTA) aggregation algorithms.

comment: The Journal is submitted to IEEE Transactions in the Internet of Things

☆ Time-to-Lie: Identifying Industrial Control System Honeypots Using the Internet Control Message Protocol

Jacob Williams, Matthew Edwards, Joseph Gardiner

The convergence of information and operational technology networks has created previously unforeseen security issues. To address these issues, both researchers and practitioners have integrated threat intelligence methods into the security operations of converged networks, with some of the most valuable tools being honeypots that imitate industrial control systems (ICS). However, the development and deployment of such honeypots is a process rich with pitfalls, which can lead to undiagnosed weaknesses in the threat intelligence being gathered. This paper presents a side-channel method of covertly identifying ICS honeypots using the time-to-live (TTL) values of target devices. We show that many ICS honeypots can be readily identified, via minimal interactions, using only basic networking tools. In a study of over 8,000 devices presenting as ICS systems, we detail how our method compares to an existing honeypot detection approach, and outline what our methodology reveals about the current population of live ICS honeypots. In demonstrating our method, this study aims to raise awareness of the viability of the TTL heuristic and the prevalence of its misconfiguration despite its presence in literature.

comment: 11 pages, 2 listings, 5 tables, 6 figures

☆ Entity-based Reinforcement Learning for Autonomous Cyber Defence CCS 2024

Isaac Symes Thompson, Alberto Caron, Chris Hicks, Vasilios Mavroudis

A significant challenge for autonomous cyber defence is ensuring a defensive agent's ability to generalise across diverse network topologies and configurations. This capability is necessary for agents to remain effective when deployed in dynamically changing environments, such as an enterprise network where devices may frequently join and leave. Standard approaches to deep reinforcement learning, where policies are parameterised using a fixed-input multi-layer perceptron (MLP) expect fixed-size observation and action spaces. In autonomous cyber defence, this makes it hard to develop agents that generalise to environments with network topologies different from those trained on, as the number of nodes affects the natural size of the observation and action spaces. To overcome this limitation, we reframe the problem of autonomous network defence using entity-based reinforcement learning, where the observation and action space of an agent are decomposed into a collection of discrete entities. This framework enables the use of policy parameterisations specialised in compositional generalisation. Namely, we train a Transformer-based policy on the Yawning Titan cyber-security simulation environment and test its generalisation capabilities across various network topologies. We demonstrate that this approach significantly outperforms an MLP-based policy on fixed networks, and has the ability for zero-shot generalisation to networks of a different size to those seen in training. These findings highlight the potential for entity-based reinforcement learning to advance the field of autonomous cyber defence by providing more generalisable policies capable of handling variations in real-world network environments.

comment: Material to appear in the proceedings of the 1st International Workshop on Autonomous Cybersecurity at ACM CCS 2024

☆ Securing Federated Learning Against Novel and Classic Backdoor Threats During Foundation Model Integration

Xiaohuan Bi, Xi Li

Federated learning (FL) enables decentralized model training while preserving privacy. Recently, integrating Foundation Models (FMs) into FL has boosted performance but also introduced a novel backdoor attack mechanism. Attackers can exploit the FM's capabilities to embed backdoors into synthetic data generated by FMs used for model fusion, subsequently infecting all client models through knowledge sharing without involvement in the long-lasting FL process. These novel attacks render existing FL backdoor defenses ineffective, as they primarily detect anomalies among client updates, which may appear uniformly malicious under this attack. Our work proposes a novel data-free defense strategy by constraining abnormal activations in the hidden feature space during model aggregation on the server. The activation constraints, optimized using synthetic data alongside FL training, mitigate the attack while barely affecting model performance, as the parameters remain untouched. Extensive experiments demonstrate its effectiveness against both novel and classic backdoor attacks, outperforming existing defenses while maintaining model performance.

☆ Differentially Private Learning Needs Better Model Initialization and Self-Distillation

Ivoline C. Ngong, Joseph P. Near, Niloofar Mireshghallah

Differentially private SGD (DPSGD) enables privacy-preserving training of language models, but often reduces utility, diversity, and linguistic quality. We introduce DPRefine, a three-phase method that initializes a model using data synthesis from a small pre-trained LM with rigorous filtering, applies DP finetuning on private data, and performs self-distillation to refine outputs. This approach significantly outperforms vanilla DPSGD, with AlpacaEval preferring DPRefine's generations in 78.4% of cases across all datasets. Our analysis reveals that DPRefine reduces linguistic errors in generated text by 84.0%, mitigating grammar and spelling errors, commonly associated with DPSGD. It also reduces inconsistencies of non-private models, such as hallucinated details and misattributed quotes. We find that small models like GPT-2 can be effective for initialization and distillation, highlighting their potential in enabling scalable and efficient deployment of privacy-preserving language.

comment: 18 pages

☆ FedGMark: Certifiably Robust Watermarking for Federated Graph Learning NeurIPS 2024

Yuxin Yang, Qiang Li, Yuan Hong, Binghui Wang

Federated graph learning (FedGL) is an emerging learning paradigm to collaboratively train graph data from various clients. However, during the development and deployment of FedGL models, they are susceptible to illegal copying and model theft. Backdoor-based watermarking is a well-known method for mitigating these attacks, as it offers ownership verification to the model owner. We take the first step to protect the ownership of FedGL models via backdoor-based watermarking. Existing techniques have challenges in achieving the goal: 1) they either cannot be directly applied or yield unsatisfactory performance; 2) they are vulnerable to watermark removal attacks; and 3) they lack of formal guarantees. To address all the challenges, we propose FedGMark, the first certified robust backdoor-based watermarking for FedGL. FedGMark leverages the unique graph structure and client information in FedGL to learn customized and diverse watermarks. It also designs a novel GL architecture that facilitates defending against both the empirical and theoretically worst-case watermark removal attacks. Extensive experiments validate the promising empirical and provable watermarking performance of FedGMark. Source code is available at: https://github.com/Yuxin104/FedGMark.

comment: This paper is accepted to NeurIPS 2024

☆ BadFair: Backdoored Fairness Attacks with Group-conditioned Triggers EMNLP 2024

Jiaqi Xue, Qian Lou, Mengxin Zheng

Attacking fairness is crucial because compromised models can introduce biased outcomes, undermining trust and amplifying inequalities in sensitive applications like hiring, healthcare, and law enforcement. This highlights the urgent need to understand how fairness mechanisms can be exploited and to develop defenses that ensure both fairness and robustness. We introduce BadFair, a novel backdoored fairness attack methodology. BadFair stealthily crafts a model that operates with accuracy and fairness under regular conditions but, when activated by certain triggers, discriminates and produces incorrect results for specific groups. This type of attack is particularly stealthy and dangerous, as it circumvents existing fairness detection methods, maintaining an appearance of fairness in normal use. Our findings reveal that BadFair achieves a more than 85% attack success rate in attacks aimed at target groups on average while only incurring a minimal accuracy loss. Moreover, it consistently exhibits a significant discrimination score, distinguishing between pre-defined target and non-target attacked groups across various datasets and models.

comment: Accepted by EMNLP 2024

♻ ☆ Feature Homomorphism -- A Cryptographic Scheme For Data Verification Under Ciphertext-Only Conditions

Huang Neng

Privacy computing involves the extensive exchange and processing of encrypted data. For the parties involved in these interactions, how to determine the consistency of exchanged data without accessing the original data, ensuring tamper resistance, non-repudiation, quality traceability, indexing, and retrieval during the use of encrypted data, which is a key topic of achieving "Data Availability versus Visibility". This paper proposes a new type of homomorphism: Feature Homomorphism, and based on this feature, introduces a cryptographic scheme for data verification under ciphertext-only conditions. The proposed scheme involves designing a group of algorithms that meet the requirements outlined in this paper, including encryption/decryption algorithms and Feature Homomorphic Algorithm. This group of algorithms not only allows for the encryption and decryption of data but also ensures that the plaintext and its corresponding ciphertext, encrypted using the specified encryption algorithm, satisfy the following property: the eigenvalue of the plaintext obtained using the Feature Homomorphic Algorithm is equal to the eigenvalue of the ciphertext obtained using the same algorithm. With this group of algorithms, it is possible to verify data consistency directly by comparing the eigenvalues of the plaintext and ciphertext without accessing the original data (i.e., under ciphertext-only conditions). This can be used for tamper resistance, non-repudiation, and quality traceability. Additionally, the eigenvalue can serve as a ciphertext index, enabling searchable encryption. This scheme completes a piece of the puzzle in homomorphic encryption. Keywords: Privacy Computing, Data Consistency, Searchable Encryption, Zero-Knowledge Proof, Feature Homomorphism

♻ ☆ KeySpace: Public Key Infrastructure Considerations in Interplanetary Networks

Joshua Smailes, Sebastian Köhler, Simon Birnbach, Martin Strohmeier, Ivan Martinovic

As satellite networks grow larger and begin to incorporate interplanetary communication, there is an increasing interest in the unsolved problem of how to approach PKI in these conditions. In this paper we explore the goals and requirements for implementing key management systems in satellite networks, focusing on megaconstellations and interplanetary networks. We design a set of standardized experiments which can be used to compare systems against one another for particular network topologies. Using these, we demonstrate that terrestrial PKI techniques are feasible in highly distributed interplanetary networks, showing that it is possible to configure PKI systems to achieve efficient low-latency connection establishment, and minimize the impact of attacks through effective revocations. We evaluate this by building the Deep Space Network Simulator (DSNS), a novel network simulator aimed at efficient simulation of large space networks. We run simulations evaluating connection establishment and key revocation under a wide range of PKI configurations. Finally, we propose and evaluate two additional configuration options: OCSP Hybrid, and the use of relay nodes as a firewall. Together these minimize the extent of the network an attacker can reach with a compromised key, and reduce the attacker's load on interplanetary relay links.

comment: 14 pages, 9 figures, 7 tables

♻ ☆ Safeguard is a Double-edged Sword: Denial-of-service Attack on Large Language Models

Qingzhao Zhang, Ziyang Xiong, Z. Morley Mao

Safety is a paramount concern of large language models (LLMs) in their open deployment. To this end, safeguard methods aim to enforce the ethical and responsible use of LLMs through safety alignment or guardrail mechanisms. However, we found that the malicious attackers could exploit false positives of safeguards, i.e., fooling the safeguard model to block safe content mistakenly, leading to a new denial-of-service (DoS) attack on LLMs. Specifically, by software or phishing attacks on user client software, attackers insert a short, seemingly innocuous adversarial prompt into to user prompt templates in configuration files; thus, this prompt appears in final user requests without visibility in the user interface and is not trivial to identify. By designing an optimization process that utilizes gradient and attention information, our attack can automatically generate seemingly safe adversarial prompts, approximately only 30 characters long, that universally block over 97\% of user requests on Llama Guard 3. The attack presents a new dimension of evaluating LLM safeguards focusing on false positives, fundamentally different from the classic jailbreak.

♻ ☆ Cyber-Physical Authentication Scheme for Secure V2G Transactions

Yunwang Chen, Yanmin Zhao, Siuming Yiu

The rapid adoption of electric vehicles (EVs) globally has catalyzed the need for robust cybersecurity measures within vehicle-to-grid (V2G) networks. As these networks are increasingly being integrated into smart charging infrastructures, they also introduce new vulnerabilities that threaten grid stability and user privacy This paper proposes a cyber-physical authentication protocol and trading smart contract tailored to plug and charge (PnC) operations within blockchain-based V2G systems. The protocol leverages advanced cryptographic techniques and blockchain to ensure secure, transparent, and tamper-proof energy transactions between EVs and charging stations. Key contributions include the development of a cyber-physical authentication method, the implementation of a smart contract framework for secure energy trading, and a detailed security and privacy analysis. The proposed protocol effectively mitigates risks such as man-in-the-middle (MitM) attacks and replay attacks while preserving user anonymity and data integrity.

♻ ☆ Gradient-based Jailbreak Images for Multimodal Fusion Models

Javier Rando, Hannah Korevaar, Erik Brinkman, Ivan Evtimov, Florian Tramèr

Augmenting language models with image inputs may enable more effective jailbreak attacks through continuous optimization, unlike text inputs that require discrete optimization. However, new multimodal fusion models tokenize all input modalities using non-differentiable functions, which hinders straightforward attacks. In this work, we introduce the notion of a tokenizer shortcut that approximates tokenization with a continuous function and enables continuous optimization. We use tokenizer shortcuts to create the first end-to-end gradient image attacks against multimodal fusion models. We evaluate our attacks on Chameleon models and obtain jailbreak images that elicit harmful information for 72.5% of prompts. Jailbreak images outperform text jailbreaks optimized with the same objective and require 3x lower compute budget to optimize 50x more input tokens. Finally, we find that representation engineering defenses, like Circuit Breakers, trained only on text attacks can effectively transfer to adversarial image inputs.

♻ ☆ Few-Shot Adversarial Prompt Learning on Vision-Language Models NeurIPS 2024

Yiwei Zhou, Xiaobo Xia, Zhiwei Lin, Bo Han, Tongliang Liu

The vulnerability of deep neural networks to imperceptible adversarial perturbations has attracted widespread attention. Inspired by the success of vision-language foundation models, previous efforts achieved zero-shot adversarial robustness by aligning adversarial visual features with text supervision. However, in practice, they are still unsatisfactory due to several issues, including heavy adaptation cost, suboptimal text supervision, and uncontrolled natural generalization capacity. In this paper, to address these issues, we propose a few-shot adversarial prompt framework where adapting input sequences with limited data makes significant adversarial robustness improvement. Specifically, we achieve this by providing adversarially correlated text supervision that is end-to-end learned from adversarial examples. We also propose a novel training objective that enhances the consistency of multi-modal features while encourages differentiated uni-modal features between natural and adversarial examples. The proposed framework gives access to learn adversarial text supervision, which provides superior cross-modal adversarial alignment and matches state-of-the-art zero-shot adversarial robustness with only 1% training data. Code is available at: https://github.com/lionel-w2/FAP.

comment: NeurIPS 2024

♻ ☆ Belt and Braces: When Federated Learning Meets Differential Privacy

Xuebin Ren, Shusen Yang, Cong Zhao, Julie McCann, Zongben Xu

Federated learning (FL) has great potential for large-scale machine learning (ML) without exposing raw data.Differential privacy (DP) is the de facto standard of privacy protection with provable guarantees.Advances in ML suggest that DP would be a perfect fit for FL with comprehensive privacy preservation. Hence, extensive efforts have been devoted to achieving practically usable FL with DP, which however is still challenging.Practitioners often not only are not fully aware of its development and categorization, but also face a hard choice between privacy and utility. Therefore, it calls for a holistic review of current advances and an investigation on the challenges and opportunities for highly usable FL systems with a DP guarantee. In this article, we first introduce the primary concepts of FL and DP, and highlight the benefits of integration. We then review the current developments by categorizing different paradigms and notions. Aiming at usable FL with DP, we present the optimization principles to seek a better tradeoff between model utility and privacy loss. Finally, we discuss future challenges in the emergent areas and relevant research topics.

comment: 10 pages, 4 figures, accepted by and to appear in Communications of the ACM (CACM)

♻ ☆ DIP-Watermark: A Double Identity Protection Method Based on Robust Adversarial Watermark

Yunming Zhang, Dengpan Ye, Caiyun Xie, Sipeng Shen, Ziyi Liu, Jiacheng Deng, Long Tang

The wide deployment of Face Recognition (FR) systems poses privacy risks. One countermeasure is adversarial attack, deceiving unauthorized malicious FR, but it also disrupts regular identity verification of trusted authorizers, exacerbating the potential threat of identity impersonation. To address this, we propose the first double identity protection scheme based on traceable adversarial watermarking, termed DIP-Watermark. DIP-Watermark employs a one-time watermark embedding to deceive unauthorized FR models and allows authorizers to perform identity verification by extracting the watermark. Specifically, we propose an information-guided adversarial attack against FR models. The encoder embeds an identity-specific watermark into the deep feature space of the carrier, guiding recognizable features of the image to deviate from the source identity. We further adopt a collaborative meta-optimization strategy compatible with sub-tasks, which regularizes the joint optimization direction of the encoder and decoder. This strategy enhances the representation of universal carrier features, mitigating multi-objective optimization conflicts in watermarking. Experiments confirm that DIP-Watermark achieves significant attack success rates and traceability accuracy on state-of-the-art FR models, exhibiting remarkable robustness that outperforms the existing privacy protection methods using adversarial attacks and deep watermarking, or simple combinations of the two. Our work potentially opens up new insights into proactive protection for FR privacy.

♻ ☆ Understanding the Effectiveness of Large Language Models in Detecting Security Vulnerabilities

Avishree Khare, Saikat Dutta, Ziyang Li, Alaia Solko-Breslin, Rajeev Alur, Mayur Naik

While automated vulnerability detection techniques have made promising progress in detecting security vulnerabilities, their scalability and applicability remain challenging. The remarkable performance of Large Language Models (LLMs), such as GPT-4 and CodeLlama, on code-related tasks has prompted recent works to explore if LLMs can be used to detect vulnerabilities. In this paper, we perform a more comprehensive study by concurrently examining a higher number of datasets, languages and LLMs, and qualitatively evaluating performance across prompts and vulnerability classes while addressing the shortcomings of existing tools. Concretely, we evaluate the effectiveness of 16 pre-trained LLMs on 5,000 code samples from five diverse security datasets. These balanced datasets encompass both synthetic and real-world projects in Java and C/C++ and cover 25 distinct vulnerability classes. Overall, LLMs across all scales and families show modest effectiveness in detecting vulnerabilities, obtaining an average accuracy of 62.8% and F1 score of 0.71 across datasets. They are significantly better at detecting vulnerabilities only requiring intra-procedural analysis, such as OS Command Injection and NULL Pointer Dereference. Moreover, they report higher accuracies on these vulnerabilities than popular static analysis tools, such as CodeQL. We find that advanced prompting strategies that involve step-by-step analysis significantly improve performance of LLMs on real-world datasets in terms of F1 score (by upto 0.18 on average). Interestingly, we observe that LLMs show promising abilities at performing parts of the analysis correctly, such as identifying vulnerability-related specifications and leveraging natural language information to understand code behavior (e.g., to check if code is sanitized). We expect our insights to guide future work on LLM-augmented vulnerability detection systems.

♻ ☆ Generative AI Security: Challenges and Countermeasures

Banghua Zhu, Norman Mu, Jiantao Jiao, David Wagner

Generative AI's expanding footprint across numerous industries has led to both excitement and increased scrutiny. This paper delves into the unique security challenges posed by Generative AI, and outlines potential research directions for managing these risks.

♻ ☆ ConfusedPilot: Confused Deputy Risks in RAG-based LLMs

Ayush RoyChowdhury, Mulong Luo, Prateek Sahu, Sarbartha Banerjee, Mohit Tiwari

Retrieval augmented generation (RAG) is a process where a large language model (LLM) retrieves useful information from a database and then generates the responses. It is becoming popular in enterprise settings for daily business operations. For example, Copilot for Microsoft 365 has accumulated millions of businesses. However, the security implications of adopting such RAG-based systems are unclear. In this paper, we introduce ConfusedPilot, a class of security vulnerabilities of RAG systems that confuse Copilot and cause integrity and confidentiality violations in its responses. First, we investigate a vulnerability that embeds malicious text in the modified prompt in RAG, corrupting the responses generated by the LLM. Second, we demonstrate a vulnerability that leaks secret data, which leverages the caching mechanism during retrieval. Third, we investigate how both vulnerabilities can be exploited to propagate misinformation within the enterprise and ultimately impact its operations, such as sales and manufacturing. We also discuss the root cause of these attacks by investigating the architecture of a RAG-based system. This study highlights the security vulnerabilities in today's RAG-based systems and proposes design guidelines to secure future RAG-based systems.